Supplementary files are available for this work. For more information about accessing
these files, follow the link from the Table of Contents to "Reading the Supplementary Files”.
Table of Contents
Go Back
CREATING
COMPETITIVE ADVANTAGE
THROUGH
DYNAMIC RESOURCE INTERACTIONS
Abhijit Mandal
PhD Candidate
Dept. of Strategic & International Management
London Business School
AMandal@london.edu
CREATING COMPETITIVE ADVANTAGE
THROUGH DYNAMIC RESOURCE INTERACTIONS
ABSTRACT
In the context of the resource-based view, some researchers have pointed out that individual
resources are not valuable by themselves, but produce value only in combination with other
resources. For example, resource flows need to accumulate over time to display
characteristics of complexity. If complex behavior manifests in characteristics of complexity,
then some fundamental questions follow: What kinds of resource combinations create value?
How should resources combine to demonstrate complex behavior? What is the role of
complex behavior in generating and sustaining competitive advantage? Our belief is that a
collection of resources that interacts within itself in specific manners to produce complex
behaviors, over time, could be a source of value to organizations. We hypothesize that the
nature of resource interactions influences the resource system’s context, whose complex
dynamics moderates the relationship between discrete resources and performance.
Employing a modeling philosophy that builds on the frameworks suggested by recent
advances towards a dynamic resource-based view in conjunction with a time-evolutionary
perspective, this paper seeks to explain the divergent performance profiles presented by two
almost identical branches of an insurance firm. Different simulations are carried out to
examine the consequences of particular sets of interactions on identical initial resource
heterogeneity. The simulations differentiate the impacts of initial resource heterogeneity and
interaction amongst resources in contributing to the complex behavior of resources. Some
conclusions are drawn about the significance of these two factors in the generation and
sustenance of competitive advantage.
CREATING COMPETITIVE ADVANTAGE
THROUGH DYNAMIC RESOURCE INTERACTIONS
INTRODUCTION
One of the most critical issues in strategic management literature is to address the question:
How is value created? Research developments in the last two decades have come up with the
resource-based view, which postulated that inter-firm differences in performance arise from
the heterogeneity of resources in firms, sustained by resource position barriers (Wemerfelt,
1984) or by imperfections in strategic factor markets (Bamey, 1986; Barney,1989).
Further query into the source of resource heterogeneity has usually produced two kinds of
answers. On one hand some have pointed to the limited supply of certain kinds of resources
(Bamey, 1991; Peteraf, 1993). On the other hand we also have explanations based on the
complexity of resources - e.g. they are causally ambiguous (Amit & Schoemaker, 1993), they
are difficult to copy precisely (Barney 1991), they need time to be copied or imitated
(Dierickx & Cool, 1989), etc.
Exponents of competencies and capabilities (e.g. Hamel & Prahalad, 1990) and others like
Grant (1991) and Black & Boal (1994) proposed that resources are not necessarily valuable
by themselves, but only in combination with other resources. Collis (1994), Barney & Zajac
(1994) and Miller & Shamsie (1996) are among those who proposed that the value of a
tesource derives from its context. For Porter (1991), value creation centers on ‘activities’ that
display characteristics of complexity. The implication is that these value-creating
characteristics derive from complex behaviors of resources that have somehow combined
together.
However, few have explained how bringing together discrete resources create complexity
characteristics, whether all kinds of complexity create value and the potential role of
managers in value creation using complexity. Even when resource combinations do create
value, is the value creation due to the number of different resources that have gone into
making them complex (detail complexity’) or is it due to the special way in which the
resources have been combined (dynamic complexity)? In the strategic management literature,
there seem to be very few explanations as to what kind of combinations of resources or what
kinds of contexts of resources create value.
If value is a function of combination and context, then we can proceed by refining this into a
proposition: Value creation can arise from the dynamics of a collection of discrete resources,
which is part of and evolves in the environment of a given context, Miller (1986, 1996)
emphasizes the inadequacy of the contingency approach for strategy research. For Porter
(1991), the cross-sectional approach to strategy needs to be supplemented by a longitudinal
approach. Together, these imply that a more dynamic approach is needed to explore the
mechanisms of value creation while adding on a temporal dimension. A beginning towards
this has been made by Dierickx & Cool (1989), who point out the role of time and
' Complexity resulting from the presence of a large number of interacting elements is ‘detail complexity’ while
complex behavior of resources (over time) that arises from the consequences of the particular way(s) in which
resources are combined is called dynamic complexity. See Sterman (p. 21, 2000) for more details.
? Ata given instant, a context is the resource network made up of the joint impacts of different initial conditions,
managerial choices and resource configurations, in addition to the discrete resources themselves.
accumulation as vital facets in bestowing characteristics of complexity and in eliciting
complex behavior’.
To study the traits of complexity and what elicits complex behavior, due consideration needs
to be given to resources as well as the causal linkages between resources; these causal
linkages between resources lead to resource interactions’. This paper takes up some of these
key aspects. Specifically, this paper explores the ‘nature of resource interactions’ among
resource sub-systems? and its role in creating competitive advantage.
The method adopted here is inspired by Rouse & Daellenbach’s research framework (1999)
for studying organizational effects by investigating in organizations. This investigation in this
paper seeks to explain the divergent performance profiles presented by two branches of an
insurance firm. The fieldwork is expressed as a model that is based on the representation of
actual resources that facilitates subsequent simulation. In the first step we set up a map (or
network) that links resources (organized in sub-systems) to other relevant resources and
performance, by specifying their dynamic behavior. In subsequent steps, by changing the
nature of interactions among resources and among sub-systems, this paper will show how the
nature of certain interactions can influence the type of stability that prevails in the resource
system. In tum, the type of stability obtained shapes the overall context, whose complex
dynamics moderates the relationship between discrete resources and performance.
The rest of the paper is organized as follows: the next section will be a review with a view to
support the scope and approach that is being adopted for this paper, followed by a section that
summarizes the basis of the model. The ensuing section displays the results of the simulations
and their analyses. The penultimate section discusses the results and its theoretical
implications. The last section concludes with the results’ implications for managers.
REVIEW: THEORY & APPROACH
In the previous section it was implied that particular patterns of resource combinations could
create value. Specifically, resource interactions - within an appropriate context - are capable
of producing, over time, complex behaviors and characteristics of complexity from the
constituent resources. If a comprehensive combination of resources that influence
performance of a firm is viewed as a system, then the resource system that elicits complex
behavior may possess some characteristics of a complex system.
For example, a small initial difference between two otherwise identical systems can end up as
a large difference in the behavior and performance of the two systems, with the passage of
time. The implication for firms in a similar setting is that a tiny difference can lead to larger
differential performance - thereby resulting in value creation and competitive advantage in a
dynamic sense, for one of the two firms. A basic question thus follows: Can there be value
creation starting from discrete resources that lack the properties of “uniqueness”?
3 Complex behavior is non-linear or dysfunctional behavior with respect to time.
‘Tf resources A, B and C are causally linked in such a manner that A affects B, B affects C and C affects A, then
resources A and B are said to interact with each other, (as also B with C and A with C).
° Resource sub-systems, for now, are described as activity systems - i.e. a collection of more elementary
resources that are functionally-oriented and whose behavior can be determined independently of the context
provided by other resources or sub-systems.
Given this property of complexity, it is natural to look away from the uniqueness of discrete
resources comprising the resource system towards other features as the source of competitive
advantage. We suggest two mechanisms to explain value creation. First, initial heterogeneity
between the resources of two systems is compounded into larger differences due to the
specific interactions that occur in the systems. Second, slightly different interactions in the
systems can tweak apart the otherwise identical initial conditions to compound them into
diverse complex resources, which is an example of complex behavior.
This paper proceeds to demonstrate the first of these two mechanisms. By holding the nature
of initial heterogeneity constant while generating complex behavior, attention is diverted to
the role of resource interactions. Specifically, the nature of resource interactions can
determine the type of stability that prevails in the resource system. The type of stability
obtained shapes the overall context, whose complex dynamics moderates the relationship
between discrete resources and performance. This idea is sketched is Figure 1.
The nature of heterogeneity in resources can be of different kinds (Penrose, 1959), as
depicted in Figure 2. She discriminated amongst resources, the productive services from
resources and the link between these two (which is influenced by the firm’s management) in
the firm’s production process. In other words, there could be a difference amongst the
resources themselves or in the way in which similar / identical resources are managed.
Prahlad and Bettis (1986) later emphasized this in pointing out the mechanism of dominant
logic as a way of understanding the impact of idiosyncratic and heterogeneous traits of
management on short-term and long-term performance (Figure 3). In this paper, initial
heterogeneity of resources originates from differences in the environment, endowments or
initial conditions. Heterogeneity in resource interactions arises from differences in the causal
structure linking the resources and from managerial policies.
It is important to point out that there have been some recent studies that have considered the
impact of interactions on organizational performance (Levinthal 1997, Rivkin 2000). In both
these studies, interaction is conceptualized as a function of detail complexity, as is the
approach to the treatment of interactions. First, it is postulated that an increase in the number
of interactions leads to more specialization. Second, complexity from interactions increases
with the number of elements that interact. In contrast, this paper focuses on the nature of
interaction that plays a role in determining the context of resources. This context, in turn,
moderates the performance outcome of the resources / resource sub-systems. Accordingly,
this paper intends to throw light upon those interactions involving particular combinations of
resources and information feedback that express complex behavior over time, rather than on
those interactions which simply involve a large number of resources.
Any research method chosen to address the above questions needs to take into account the
role of resource configurations, managerial choices and initial conditions to accommodate a
large number of variables. The evolutionary underpinnings of Nelson & Winter (1982), Teece
et al (1997) and Levinthal & Myatt (1994) hold promise in portraying the dynamics of the
resources and interactions involved. Noda & Collis (2001), in a longitudinal study, show the
different kinds of feedback and the dynamic interplay among market, competitive and
organizational forces that guide the evolution of heterogeneity of firms in an industry®. The
® Due to a lack of quantification, Noda & Collis (2001) fail to show how the various forces interact with each
other and how they influence the context in a dynamic fashion. It prevents a detailed understanding of the
changing relative importance of these forces and the consequent implications for resources towards producing
competitive advantage.
evolutionary approach calls our attention to three characteristics of major importance. First,
instead of the maximizing behavior of managers assumed by the traditional RBV, this
approach adopts ‘satisficing’ in the context of bounded rationality as the dominant mode of
behavior, from the behavioralist paradigm (Cyert & March, 1963). Second, in contrast to the
papers on interactions reviewed above, the proposed model is built around real business
structures that correspond to the facts of the industry, the actual resources and their config-
urations; it is similar to the history-friendly modeling approach by Malerba, Nelson, Orsenigo
and Winter (1999, 2001). It specifies the causal relationships among the various resources
and the processes involved in the transformation of resources over time. Portraying the causal
and the process structures is essential to representing the network of resources and their time-
evolution’ such that it is able to reproduce the actual behavior of the firm(s) concemed.
The third major characteristic, which follows from the causal structure mentioned above, is
that the proposed model will be based on a non-equilibrium approach. Typically, equilibrium
models are limited to conditions of equilibrium and do not necessarily portray what happens
on the path to equilibrium®, They use mechanisms of rent appropriation to link the initial
heterogeneity in resources to the sustenance of rents (a proxy for competitive advantage).
In contrast to equilibrium models, non-equilibrium models are more robust and complex.
They provide an opportunity to clearly delve into the dynamics of resources - specifically,
the sense of how resources and capabilities evolve over time. Researchers like Porter (1991,
1996), Bromiley (1993), Collis & Montgomery (1995) and Priem & Butler (2001) have
highlighted the importance of this specification as it permits better operationalization to
examine the propositions of the resource-based perspective. In addition, these models can
separate the effects of various factors on static and dynamic bases. This paper explores initial
heterogeneity and interaction among resources in the generation of differential performance
(which results in competitive advantage).
Mahoney & Pandian (Footnote 6, 1992) suggest that resources could be denoted as stocks
and productive services from resources could be represented as flows. Dierickx and Cool
(1989), who introduced resource accumulation into strategic management literature, define
competitive advantage as a function of the levels of those asset stocks that measure
performance variables. Other researchers e.g. Decarolis & Deeds (1999) also conceptualize
relevant kinds of resources into stocks and flows. However, they do not elaborate on how an
initial heterogeneity between two firms will or will not result in a competitive advantage.
With a view to integrate the resource-based, evolutionary and accumulation perspectives
(represented in Figure 4), we adopt a system dynamics methodology to model a dynamic
resource system in order to convey our objectives for this paper.
An important characteristic of such a dynamic resource system is that it classifies all
resources into three kinds. First, there are those that do not vary with time, denoted as
constants. Next, there are resources that need time to adjust from their previous value,
because of the inertia of resource accumulation - denoted as resource stocks. Third, there are
resource flows to and from resource stocks. These resource flows are a function of the level
” Here, time-evolution is to be interpreted simply as transformation that results over the passage of time, rather
than those interpretations associated with ecology. The modeling method adopted is not that adept in handling
discrete events or events whose impact has not been conceptualized.
® Some researchers have utilized equilibrium models (Lippman and Rumelt, 1982; Oi, 1983) to explain the
persistence of some given heterogeneity. But these models do not lay out the process that resources go through,
on their way to equilibrium.
FIGURE 1-A Mechanism where
Context moderates Performance
Nature of interaction
(determined by policies,
sub-systems or structure)
TYPE OF OTHER OTHER
STABILITY FACTOR FACTOR
(stable, unstable,
neutral)
CONTEXT
DYNAMICS
Resources Performance
OF RESOURCES
FIGURE 2 - Ideas on
Competitive Advantage
Resource “Unique Resources are
Based - the key to achieving
View Strategic Advantage”
“A firm is a bundle of (Penrose’s
productive resources” interpretation)
* resources
* productive services
* relationship between
resources
and productive services
Differences in firm
performance arise due to:
a) Possession of different
resources
b) Resources are managed in
a different way
FIGURE 3 - Ideas on
Competitive Advantage
Competence “Unique Competencies are
Based - the key to achieving
View Strategic Advantage”
Differences in
Dominant Logic of
Management
Differences
in resource
mgmt.
Different
processes
Different Distinc
Competencies
Different
Organizational
Routines
Different Spec ific
Assets
FIGURE 4 - Ideas on
Competitive Advantage
Resource “Differential Accumulation of Key
Accumulation - Resources overtime achieve
View Strategic Advantage”
A firm
consists of
stocks &
flows
Dierickx + Cool’s interpretation
Critical
Resource
Firm Specific
Components
Stocks + Flows
rs : Different Differential
/° Different Distinct. Combinations} | Levelsof
“ Competencies“, Stocks
Different
Organizational
Routines } E cr
Z Differential ompetitive
Diffe rene Speciic, Performance Advantage
~ Assets wf
of the resource stocks and are capable of varying relatively instantaneously. These concepts,
including that of accumulation, assimilation, information flows and material flows are
explored in more detail in the system dynamics literature (Forrester, 1961; Sterman, 2000;
Morecroft, 1994; Morecroft, 2000).
Probably the most outstanding feature of this method of simulation is that the model
specification requires an accurate mathematical description of the activity systems prevailing
in the firm. By specifying how the transformation of resources in the activity systems affects
each other (with the help of differential equations), there is a clear ex-ante link between
resources, resource management and rent appropriation. The causal links are essentially the
building blocks of the dynamics of strategy. Moreover, they help to determine the input of
management policies towards performance and bring to life Penrose’s (1959) idea that
management is a “dynamic interactive process”. It addresses Porter’s plea (1991) to explain
how resources and capabilities evolve over time.
An important characteristic is that the method facilitates the impact of both kinds of
feedback. Goal-seeking behavior, or negative feedback, has long been recognized as an
important dynamic in the management of resources (‘homeostasis’) - particularly in
organizational theory. Various authors (Arthur, 1989, 1990; Levinthal and Myatt, 1994) have
also highlighted positive feedback. The type of feedback indicates the behavior and the
overall direction of resource accumulation in the loop. If the organization is taken as a whole,
there certainly exists a possibility of more than one type of feedback as well as changing
types of feedback. This may make it difficult for human beings to intuitively guess the final
outcome of the dynamic consequences of the time-evolution of the relevant resources.
Another major advantage of this method is its ability to quantitatively combine resource
heterogeneity arising from resource endowments, from the manner in which information is
processed and from actions taken about resources (management policies). Quantification of
the processes studied during fieldwork allows this method to improve upon the shortcomings
pointed out earlier in the approach adopted by Noda & Collis (2001).
A potentially significant application is to examine and verify the properties of resources
specified by Bamey (1986, 1991) and Peteraf (1993) in their exposition of the resource-based
view - particularly the conditions deemed necessary to generate and sustain competitive
advantage. In other words, is it possible for resources without special properties to generate
competitive advantage? Can competitive advantage arise from the interactions among
resources as a function of the time-evolution of the resource network rather than the discrete
resources themselves?
MODEL STRUCTURE
This section summarizes the model and simulation objectives. The philosophy that leads to
the model's construction is inspired by Rouse & Daellenbach (1999), who lay out a research
framework for studying organizational effects by investigating in organizations. We take
their suggestion about the selection process one step further by investigating a phenomenon
that involves two branches of a major insurance firm in the United Kingdom. These branches
were part of the same insurance group and were selling the same portfolio of products. They
were similar in resource structure and management policies in almost all aspects. In addition,
these branches were in markets that required similar levels of expertise from their manpower.
As expected, these branches initially had comparable levels of performance, but later there
was a divergence. It is the cause of this divergence in the performance level that we seek to
investigate through modeling.
Appendix 1 presents a short background of the insurance industry and a brief description of
the typical insurance firm in terms of its resources and activities performed. The agency (i.e.
sales) department affords the greatest flexibility to firm management to establish a
competitive advantage through differentiation and productivity’. Management decides what
kind of agents (sales persons) to hire, how much training they should get, how to train them,
where to spread its agents and how to identify and retain/promote its star performers. This
department is responsible for the entry of new money streams, to maintain the ‘going
concem’ status. It is also the largest cost item that can actively be managed in the business
plan of the insurance company. For these reasons this paper focuses on the agency
department of firms in the insurance industry.
Insurance industries have their own particular measures of profitability. These are mentioned
in Appendix 2. Trends in this industry are towards distribution through direct marketing,
specialization of service required by policyholders and a significant reduction in commission
based compensation. The new sales-force has to be quick at learning how to tackle issues
specific to their clients. Note the increase in the importance of agency management, as its
role becomes more critical in delivering a competitive advantage to the insurance firm.
Since we will focus on the agency department, we will henceforth denote this department as a
firm. A more detailed description of the activities for this kind of firm, along with the
performance measure adopted here, is presented in A ppendix 3. The activities are modeled as
four sub-systems: the headcount sub-system, the skill sub-system, the productivity sub-
system and the compensation sub-system. The headcount sub-system is the collection of
processes directly concemed with the management of agent headcount, which include hiring,
firing and promotions. The overall rates of agents moving into or leaving the firm and the
number of agents have an important impact on the dynamics of the skill pool of the agents.
The skill sub-system is about the management of agents’ sales skills. For purposes of
simplification, multiple dimensions of sales skill have been collapsed into one dimension,
which is based on the years of sales experience possessed by an agent. Even though the
measure of this skill is not very tangible, it is one of the most important drivers of
performance in the industry. The productivity sub-system highlights the productivity and
tumover of the sales agents and the resultant dynamic impact on the product portfolio. The
compensation sub-system models the mechanics of fixed and variable compensation
(commission), for agents and managers. It specifies how the level of skills, the lapse rate and
the quit rate of the agents affect compensation and, in tum, how the compensation level
affects the same three variables.
These sub-systems are explained in more detail in A ppendix 4, with the help of diagrams.
The sub-systems that model the various processes addressing managerial headcount, agent
learning, managerial learning, managerial responsibilities, managerial time constraints, the
comparison of performance and the reaction to the comparison by managers, have all been
excluded in order to simplify the model to a significant extent and keep it within the scope of
° Another way in which firms in the industry differentiate themselves is through the pattem of ownership of the
equity structure of the firm. However, this makes an impact only during exceptional events in the firm’s history.
this paper. The discussion section reviews some assumptions behind the managerial behavior
portrayed in the sub-systems in light of the actual behavior displayed by managers.
Following the approach suggested by Rouse & Daellenbach (1999), the cause and effect
relationships describing the processes in the model were derived from interviews with five
experienced managers from the industry and with other industry consultant experts, totaling
60 hours. Further secondary data about the major insurance firm and the insurance industry
was also used to enhance information obtained from the interviews. These relationships are
now being verified with published research. Industry experts have validated the behavior of
the model to substantiate the link between its structure and the working of the field setting.
SIMULATION: RESULTS AND ANALYSES
We recollect here the objectives of the paper that we have mentioned in passing. The first is
to demonstrate how resource interactions give rise to complex dynamic outcomes and that
different kinds of initial conditions set off different time-evolutionary paths for resources. In
effect, these interactions create a context that moderates the link amid resources and.
performance and thus results in different trajectories for performance. The second is to
impress upon readers that it is possible to differentiate between the roles played by resource
heterogeneity and interactions amongst resources, in an outcome that shows differential
performance. The third is to contrast the properties of resources specified by Bamey (1991)
and Peteraf (1993) to sustain competitive advantage with the characteristics of resources and
their interaction mechanisms that produce differential performance; these characteristics and
mechanisms are derived from the exposition of the model in this paper.
To illuminate these points, there will be a series of simulations progressively involving more
complex interactions and sub-systems'”. In the first simulation, managers have perfect
information and control about holding onto their initial competitive advantage; there should
be an ideal outcome. The next is carried out under conditions of absence of / partial
information and control; one expects the initial advantage to erode. The last simulation
integrates the compensation sub-system. In each simulation we show the time-evolution of
three very similar firms - Alpha, Beta and Gamma. Within a simulation, these firms differ
from each other only in initial heterogeneity while possessing the same causal structure; they
follow the same policies. Across the simulations this initial heterogeneity that distinguishes
the three firms is replicated in a consistent manner. The intention is to help the reader
separate the effects of resource heterogeneity and interactions amongst resources in the result.
Simulation 1 - Perfect Information and Management Control
We take a relatively simplified version of the firm independent of many of the other sub-
systems that usually form an integral part of it. Imagine a firm composed of only the
headcount, skill and productivity sub-systems. The composite diagram of the network of
resources (or causal structure) is depicted in Figure 5. In this firm, consider an ideal situation
where managers have perfect information about the ability of their agents and have attained a
perfect ability to retain their superstar agents (i.e. those with extra-ordinary skill) as
salespersons. In addition, they do not care about the skill level of the other agents who leave
the firm or of those who are promoted.
1° This style of adding on partial models and interactions to examine qualitative pattems of behavior is discussed
further in Morecroft (1984, 1985) and Sterman (2000). It is related to the dialectic inquiry method of Mason
(1969) and Mitroff et al (1982).
10
The (causal) structure that corresponds to this ideal equates the skill level of the agents in flux
(those entering, quitting and promoted) to the average level of the skills in the market - and
thus, nothing to do with the average skill level of the firm. It is shown in Figure 5 by the thick
arrows from Skill of Agent Quits and Skill of Agents Promoted to Lost Skill from Agent Quits
and Lost Skill from Agents Promoted respectively. There are no arrows from Skill per Agent
to Lost Skill from Agent Quits and Lost Skill from Agents Promoted.
As mentioned before, there are three firms - the base case (Beta) and two additional cases
(Alpha and Gamma) - that are identical except for the initial heterogeneity. All have 100
agents each at the start of the simulation. Beta has 300 equivalent years of experience in sales
skills (Agent Sales Skills) implying an initial value of Skill per Agent of 3 equivalent years of
experience per agent. The initial heterogeneity is introduced through Alpha and Gamma; they
have 330 and 270 equivalent years of experience in sales skills respectively. Standard Agent
Skill at Hire, Relative Skill of Quits and Relative Skill of Promotions are all set to 3 equivalent
years of experience, which is equal to the average levels of the skills in the market.
(Standard) Agent Quit Rate is set at 0.20 per year while Agents Promoted is set at 0.025 per
year, The values assumed are within the range prevalent in the firms that were studied'!.
Figure 6 shows the behavior of these ideal firms over ten years. The final performance levels
remain at their initial levels for Alpha, Beta and Gamma. If performance is measured in Skill
per Agent, then the firms have remained steady at 2.7, 3 and 3.3 years. Similarly, the other
measure, Profitability, is also steady at 73, 75 and 77 units respectively. The amount of Agent
Quits, Agents, Agents Promoted and Agents Hired remains equal and steady for all the three.
In this simulation, the headcount policy is simply to replace those who quit (Agent Quits) and
those who are promoted (Agents Promoted). As the number of such agents is proportional
only to the existing number of agents, the total amount of agents in the three firms remains
the same for the duration of the simulation. So does the number of Agents Promoted, Agent
Quits and thus, the hired (Agent Hires). When it comes to agent skills (Agent Sales Skill),
management retains the extra-ordinary performers as set out by the definition of the situation.
Those leaving and being promoted, not subject to any skill restrictions, exit (and enter) at a
skill level equal to the average of the market. Consequently, the skill level of the firms
remains constant at their initial levels, as observed. The same lack of dynamics holds for
Profitability.
This simulation demonstrates that when managers are capable of holding on to their scarce
and valuable resources, they can sustain their initial competitive advantage.
Simulation 2 - Minimal Information and Management Control
In contrast to the previous simulation, here we have another stylized situation where
managers have minimal information about the ability of their agents and no control over
which agents leave or are promoted. This implies a lack of ability to retain their superstar
agents. Accordingly, the skill level of the agents in exit is set to the average levels of the
skills in the firm. It is shown in Figure 7 by the dotted arrows from Skill per Agent to Lost
Skill from (Agent) Quits and Lost Skill from (Agent) Promotions. These Lost Skills are no
" Very few of the values used have been adjusted to clarify and communicate the issues that prompted the
simulation. Nevertheless, all the values being used are within the range of values that were prevalent in the
firms.
11
FIGURE 5 - Model for Simulation 1
fi
lost skill from
. agent quits ¢
Noe lost skill from al t
4+ agents promoted per agen
skill of
q ee oi ‘ agent quits
+
added skill
at hire
4
standard agent
skill at hire
+ + +
skill of agent quits = A---2 B An on inA
causes B to increase
skill of agents promoted =
standard agent skill at hire = A--- B Anincrease in A
average market level of skills = causes B to decrease
3 years of experience Reinforcing loop
1 of
Balancing loop
12
FIGURE 6 - Results of Simulation 1
Graph for skill per agent
3.5]
2.5.
Time (year)
Alpha: 3.3 equivalent years of experience. — —
Beta: 3.0 equivalent years of experience
Gamma: 2.7 equivalentyears of experence— — — — — — — — — — —-
Units: equivalent years of experience (above) and chrrensioness (aio)
Graph for profitability
80
78)
76)
74|
72)
70
0
Time (year)
FIGURE 7 - Model for Simulation 2
=
is -
+ agent
i ge
hiresy+ agents quits
promot
Vv, lost oll from.
lost. ae sil it aan
agents promoted Per ager
a
a
4
jents”
+ ~ Sales Skill
added skill Al
standard agent
skill at hire
13
longer a function of the average levels of the skills in the market. This change from the
previous simulation introduces the first significant interaction amongst the resources in play;
it may be almost conceptualized as a different policy in implementation.
How would the conditions that are specified above influence the performance of the firms?
Since introducing this interaction brings into play two more balancing loops than the previous
simulation, one expects a performance profile that is ‘more stable’. However, there still
remains the question, more stable with respect to what? Recall that in the last simulation, all
the three firms had very stable and consistent performance profiles. As before, we proceed
with the three firms, identical to Simulation 1 with respect to their initial values. Relative
Skill of Quits and Relative Skill of Promotions are set to an absolute value of 1, a clear
indication of management powerlessness with respect to the market. Agent Quit Rate is set at
0.20 per year while Agents Promoted is set at 0.025 per year, as in Simulation 1.
Figure 8 shows the dynamic behavior for this stylized version over ten years. It shows that
Beta remained steady at 3 years in terms of Skill per Agent and at 75 units in terms of
Profitability. Note that the final performance levels of Alpha and Gamma converge to that of
the base case whether performance is measured as Skill per Agent or as Profitability. The
number of Agents, Agent Quits, Agents Promoted and Agents Hired remains invariant over
time and equal for all the three firms.
Although the headcount policy is very similar to the previous simulation, the skill retention
policy is quite different. Management is almost powerless to control who leaves the firms. If
agents leave the firms at random, it is equivalent to the situation where the skills of those
leaving, at a given point, are the same as the average skill level (Skill per Agent) of the firms
at that instant. This relation with the instantaneous average skill level indicates the presence
of the new significant interaction whose effect is highlighted. In terms of causal structure,
additional balancing loops (marked in the figure) connecting Agent Sales Skill, Skill per
Agent and Lost Skill from Agent Quits / Lost Skill from Agent Promoted are active.
These additional balancing loops influence the rate of accumulation and the time-evolution of
the firms’ resources by modifying the interaction context that was pointed out earlier. The
combined effect of the new balancing loops and other existing loops manifests through a
mechanism typical of the dynamics of balance. When agents with above average or below
average skills leave the firms, their substitutes are hires whose skill level (Level of Hired
Agents) is at the average level of the market. As this mechanism plays out over time, the skill
level in both Alpha and Gamma converges to that of the market and that of Beta. Profitability
is directly linked to Skill per Agent; therefore it exhibits similar paths.
Modification to Imperfect Management Information and Control
To paint a scenario that is more realistic than the one immediately above, consider a situation
where managers have imperfect information regarding which agents leave or are promoted. It
continues to imply a lack of significant ability on the part of management to retain their
superstar agents. The corresponding causal structure is repeated in Figure 9 by dotted arrows.
The difference to the causal structure is that management exercises some control regarding
who is promoted or who quits because of changes in the values of Relative Skill of Quits (set
to 0.90) and Relative Skill of Promotions (set to 1.80). These are shown as thick arrows.
A value of Relative Skill of Quits lower than one implies that only worse than average agents
quit; it increases the skill level of the remaining agents. On the contrary a value of Relative
14
Skill of Promotions higher than one implies that only better than average agents advance; it
lowers the skill level of the remaining agents. While these abilities are more in line with what
management would desire, their effects are contradictory. These modifications do not really
affect the impact of the significant interaction that was presented in the last simulation. The
performance trajectory of Alpha and Gamma converge to that of Beta, which is steady at 3
years in terms of Skill per Agent and at 75 units in terms of Profitability. All facets of the
outcome are similar to the previous simulation, where managers have minimal information.
The convergence proves that the dynamics of the balancing loop(s) dominates the time-
evolution of the resource context. This simulation demonstrates that when managers are
incapable of holding on to their scarce and valuable resources, their initial competitive
advantage gets eroded.
Simulation 3 - The Complete Model
In this simulation we integrate the hitherto neglected compensation sub-system with the
imperfect management information and control scenario simulated above. The additional
interaction that is now being facilitated comes not from a simple change in abilities but from
the complex behavior of the compensation sub-system itself (explored in A ppendix 5). From
there we concluded that although there aren’t any policy levers that managers consider to be
practical, the sub-system functions in such a manner that deviations from the steady-state
input value of compensation results in more deviation before stabilizing.
Is the integration of the compensation sub-system and its ensuing additions to the nature of
interactions likely to result in a substantially different outcome? Given that this aspect brings
into play more reinforcing loops than the previous simulation, one expects a performance
profile that could perhaps be ‘less stable’. Does this mean that the convergence seen in
Simulation 2 will revert back to the steady pattern that was the outcome of Simulation 1 or
even beyond? The precise outcome is somewhat difficult to predict by intuition.
Figure 10 shows the complete diagram of the model in use for this simulation. Note the
interactions with the previous simulation model are transmitted through the causal links
shown by the thick arrows in the diagram. As usual, there are the firms Alpha, Beta and
Gamma with their initial details identical to Simulations 1 and 2. Values related to the
compensation sub-system have been explained in A ppendix 5.
Figure 11 shows the dynamic behavior of the firms for twenty years. The firm Beta repeats a
steady performance profile that is linear and flat. As before, the outcome is 3 years in terms
of Skill per Agent and 75 units in terms of Profitability. However, Alpha and Gamma diverge
away from the trajectory of Beta, in contrast to the convergence that was observed in
Simulation 2. Further, at any point in time, there is a difference among the three firms in the
number of Agent Quits and Agents Hired, but not in the total quantity of Agents.
Recall that the initial values of the three firms and the discretionary policies followed in
Simulation 3 are the same as in Simulation 2, wherever applicable. To explain the seemingly
contradictory behavior we trace the impact of a key variable - Skill per Agent - on the newly
formed loops. These reinforcing loops have been created by integrating the compensation
sub-system with the other loops that existed in the causal structure of the resources prior to
this simulation. We follow the case of firm Alpha where the initial value of Relative Agent
Skill is greater than the steady state value of 1.
15
FIGURE 8 - Results of Simulation 2
Graph for Skill per Agent
Alpha: 33 equvalentyears of experience = == === == -- === -
Bel: 3.0 equivalent years of experience
Gamma: 2. equvalent years of experience — — —_—_ _—__—_.
Units: equivelent years of experience (above) and dirrersioriess (blow)
Graph for profitability
80)
78)
76 iis a
Time (year)
FIGURE 9 - Model for
Simulation 2 (extension)
Ps
% its
mote
Telative skill +
of --
promotions + ag a from +
— skill agent quits
lost. ie from i agent
= ma promot
relative skill
in a of quits
- pe Skill
added skill _—%.
+
standard agent
skill at hire
16
FIGURE 10 - Model for Simulation 3
of promotion t
n 1 lest ski from
a—— skill ‘agent quitst
lost skil from i
+* gents promoted Egent
fun
x
relative skill
of quits
la
variable
compensation
Helalive
compensation
level of
hired
standard agent agents 7°
skill at hire
FIGURE 11 - Results of Simulation 3
Graph for Skill per Agent
Time (year)
Alpha: 33 equivalent years of experience — — —
Beta: 3:0 equivalent years of experience,
Gamma: 2.7 equivalent years of experience — — — — — — — ee
Units: exivelent years of experience (above) and dirrensioriess (blow)
Graph for profitability
Time (year)
17
Appendix 5 demonstrates how an increase in the value of Relative Agent Skill beyond its
steady state value leads to a decline in the value of Agent Quit Rate (with some delay)
through a delayed increase in Recent Relative Compensation. In the situation of Simulation 3,
this becomes an interaction effect - rather, a feedback effect; Agent Quit Rate is no longer a
constant as it was in Simulation 2, but has now become an indirect function of Skill per
Agent. If one traces the reinforcing loops from Skill per Agent to Sales Productivity to the
compensation sub-system to Agent Quit Rate to the manpower sub-system, the net effect is to
increase the accumulation of skills in the stock Agents’ Sales Skill with respect to the
accumulation of Agents; each iteration reinforces the relative increase in Skill per Agent and
Relative Agent Skill.
In addition, Appendix 5 demonstrates how the above increase in Relative Agent Skill leads to
an increase in the value of Level of Hired Agents. Again, in this context it becomes a feed-
back effect; Level of Hired Agents is no longer a constant but an indirect function of Skill per
Agent. The corresponding reinforcing loops (in Figure 10) are from Skill per Agent to Sales
Productivity to the compensation sub-system to Level of Hired Agents to the manpower sub-
system. The net effect of each iteration is exactly as spelt out in the previous paragraph.
The accumulation and feedback processes cited in the loops influence the time-evolution of
the firms’ resources by continuously modifying the existing interaction context. Their
combined effect manifests through an accumulation mechanism that opposes and over-
compensates the dynamic balance mechanism activated from Simulation 2. The implication is
that a firm which has an initial positive displacement of skills with respect to the level
required by the market will continue to attract more and more skilled personnel, bringing it
more prosperity.
For the firm Gamma, where the initial value of Relative Agent Skill is less than the steady
state value of 1, the explanation just described works in a symmetrically reverse manner
implying steadily decreasing performance. This inherent tendency of the causal structure to
move away from the market equilibrium produces the overall divergence effect. The
dynamics of reinforcing loops create instability; the context that arises from it demonstrates a
decisive impact on differential performance, which results in competitive advantage.
The following, Table 1, summarizes the different simulations and the outcomes.
Simulation | Existing New Interactions Performance | Comments
Interactions Profile
1 Minor - Stable, flat Heterogeneity
sustained
2 Asin1 Lost Skill from Agents Convergent | Heterogeneity
Promoted = fn.(Skill per erodes with
Agent, Relative Skill of time
Promotions)
Lost Skill from Agents Quits
=fn(Skill per Agent, Relative
kill of Quits)
3 Asin2 Agent Quit Rate =tn.(Recent | Divergent Heterogeneity
Relative Compensation) amplifies with
Level of Hired Agents = time
fin.(Recent Relative
Compensation)
18
DISCUSSION
This section discusses the theoretical implications of the simulation results. So far, the source
of heterogeneity in the firms Alpha and Gamma has been taken for granted. There are many
ways of arriving at this initial situation - e.g. a difference in policies between firms
propagated for some time, even with no other initial heterogeneity. However, it is beyond the
scope of this paper to address the dynamics that lead to the initial situation presented here; it
is addressed in further research. The lack of managers’ response to declining performance in
the short and medium term is reviewed here, in light of the sub-systems shown in the model.
In the model, one of the key determinants of performance is Skill per Agent. When managers
find a decrease in Skill per Agent, theoretically they have three kinds of intervention modes:
altering the value of Relative Skill of Promotions and/or Relative Skill of Quits, changing the
Agent Promotion Rate and intervening in the compensation sub-system. Interviews and
accounts from industry experts confirm that managers are too busy in the day-to-day
responsibilities of recruiting agents and coaching them, partly because of job requirements
and partly because they want to justify the agents they chose. To a large extent, managers
effectively find it difficult to monitor the churn rate of agents, let alone consciously monitor
the fluctuations and the trajectory of Skill per Agent; typically they are ignorant about it until
too late. Even if a minority of managers is aware of the situation, there are other difficulties
about intervening in the system using this information; these are summarized below.
In the model presented, managers would seek to decrease the values of Relative Skill of Quits
and Relative Skill of Promotions to the minimum to improve the average skill level of the
agents involved in sales. This implies tracking and analyzing the performance of individual
agents. Since it involves a lot of additional effort for managers and requires making biased
judgments affecting agents’ careers, they would rather go with simple and robust policies that
provide greater transparency for all, despite compromising on efficiency. Similarly, managers
do not manipulate the compensation sub-system away from standard industry practices for
the majority of the agents as per industry norms; they focus attention on retaining outstanding
performers through non-financial incentives, though this is not always successful. The ratio
of managers to agents is decided and monitored by industry regulators; consequently there is
no room for managers to change the value of Agent Promotion Rate to their convenience.
The role of the firms Alpha and Gamma in the series of simulations was to control for the
nature of initial heterogeneity. Given the systematic changes in resource interactions and
corresponding performance profiles, this will assist readers to differentiate between the
impact of resource heterogeneity and interaction amongst resources as well as estimate the
significance of these factors in the generation and sustenance of competitive advantage.
If the outcome of the first simulation is evaluated in terms of competitive advantage, the
firms demonstrate perfect sustainability of their initial competitive advantage / disadvantage.
Proponents of RBV e.g. Barney (1986, 1989, 1991), Peteraf (1993) etc. have specified certain
properties that resources must possess to ensure the sustenance of competitive advantage. We
compare the specified properties with the properties of resources inherent in the simulation.
According to Barney (1991), the resource must be valuable, rare, inimitable and non-
substitutable to sustain competitive advantage. A gents with outstanding skills are
undoubtedly of value. In the first simulation, they are rare in the sense that they are not
regularly recruited from the market, and inimitable as they cannot be developed in-house -
19
but true to the meaning of initial heterogeneity, they exist at the initial point in time.
Together, these imply that the condition of non-substitutability holds. Peteraf (1993) holds
that there must be ex-post and ex-ante limits to competition for the critical resource, which
must also be superior and imperfectly mobile. The first and the last conditions are satisfied
since the agents with outstanding abilities fulfil Ghemawat’s ideal of sticky resources (1992).
They are superior in performance and available in limited supply, as shown. The condition
prescribing limits to ex-ante competition also holds. Since the conditions specified by Bamey
and Peteraf are met in full, perfect sustainability is predicted, and it follows.
The resource stock Agents possesses the characteristics of a resource that is influenced by the
interaction of other resources (Agents Promoted, Agent Quits and Agents Hired). However, in
the first simulation, the actual values of these resources are such that no complex behavior is
observed. Hence this model serves as a sort of a base to build upon, and with which we could
contrast the other simulations.
The firm in the second simulation fails to sustain its initial competitive advantage due to the
introduction of a new kind of interaction. The type of interaction that interests us is that the
skills of those agents who leave the firm (Lost Skill from Agent Quits, Lost Skill from Agents
Promoted) are proportional to the instantaneous average skill level (Skill per Agent), which is
a function of the individual agent’s skill, a valuable resource. An examination of the
interaction effect on this valuable resource shows that the resource is neither rare nor
inimitable, violating Barney’s conditions. Similarly, the mobility of agents violates one of
Peteraf’s requirements. Since some of their specified conditions are not met, the theory
predicts non-sustainability of the initial competitive advantage; convergence confirms that.
The nature of this kind of interaction also creates a context that enhances stability. Compared
to the first simulation, this new context, despite similar initial heterogeneity, has sufficiently
changed the dynamics of the involved resources to bring about a change in the profile of the
firms’ performance trajectories. Hence we claim that certain interactions moderate the
relationship between resources and performance, through the dynamics of resources.
In the third and last simulation, the performance profile of the firms shows divergence due to
interaction with the compensation sub-system. Management persists with the policies that
brought convergence in the second simulation. Their interactions continue to violate the
requirements specified by Barney and Peteraf. However, these interactions are now enmeshed
in another kind of interaction-driven context that reverses the direction of mobility of the
valuable resource pointed out above; instead of convergence this leads to the observed
divergence and differential performance. From a dynamic point of view, it seems the
conditions that resources must fulfil to sustain competitive advantage are over-specified.
There arises the distinct possibility that competitive advantage, which results from
differential performance, might be created from a suitable combination of ordinary resources
without satisfying all the conditions laid out by Barney or Peteraf.
From the perspective of causal structure, interactions with the compensation sub-system
simply means that the flow of resources to the stock Agent Sales Skill has put that stock into
an accumulation mode that orients it towards instability rather than stability. The context
arising from this has overhauled the dynamics of the involved resources to reverse the profile
of performance trajectories.
20
With respect to Simulation 1, Simulations 2 and 3 have shown how resource interactions can
bring about complex behaviors, demonstrating the creation of competitive advantage through
dynamic complexity. The contrast between Simulation 2 and Simulation 3 shows that while
resource heterogeneity may be a necessary condition for differential performance, it is the
nature of interactions that determine how far the heterogeneity will continue. The outcome of
the last simulation provides support for our claim that the nature of interactions is a key
ingredient in creating the context for the interaction of resources that moderates the
relationship between resources and performance, through the dynamics of resources.
CONCLUSION
Having demonstrated and discussed the objectives that were set for this paper, this section
attends to some related issues for research and their implications for managers in enhancing
competitive advantage for their organizations. One of the important characteristics of the
method used is the disaggregation of resource relationships and the specification of
processes; researchers following the frameworks of Rouse & Daellenbach (1999) and Noda
& Collis (2001) can go beyond producing a case-study to provide managers with levers and
mechanisms that might enhance success in past and future environments. In addition,
managers can quantitatively examine the range of operational freedom that these levers and
mechanisms provide, apart from being able to compare their relative ease of use and the
speed and consequences of their impact.
Further, the method answers Bromiley’s call (1993) for a better operationalization of the
propositions of the resource-based view, in a manner such that it can better overcome the
charge of tautology (Priem & Butler, 2001). By specifying causal linkages with intermediate
variables, it provides synthetic instead of analytic statements (Powell, 2001) that facilitates
examination of the propositions of the resource-based view in a more exacting manner.
The method may not always help pinpoint the singular source of competitive advantage.
However, if managers are convinced that the manner of combining resources is important and
can be a source of competitive advantage in its own right, then the method does facilitate the
identification of those critical resource stock accumulations that are vital for competitive
success. Based on this, managers can derive actionable implications to attain competitive
advantage for diverse circumstances. Exploring the consequences of proposed actions
through a simulation helps to arrive at policies that may be more robust. Further, it can be
used to accurately identify - even quantify - the relative contributions of the various sources
of differential performance (e.g. initial heterogeneity, interactions among resources) when
one examines how initial heterogeneity is propagated through time.
From Simulation 2, one might infer that the endowment of resources, or the initial position,
makes no difference or plays no role in the ultimate outcome. In contrast, Simulation 3 shows
that the initial position makes a critical difference in the ultimate outcome. This shows that
the importance of certain resources (in this case, Skill per Agent) can diminish or increase
with respect to time. Moreover, what was a resource with little influence on performance in
Simulation 2 has become of critical importance in Simulation 3 - implying that the
importance of resources can change with the policies being followed by management and the
context allowed for the time-evolution of resources. This conclusion suggests that more
attention should be paid to the role of path dependence in creating value for organizations.
21
According to the resource-based view, mobility of resources is an important factor for
competitive advantage. Peteraf (1993) takes a protective view of mobility: for her, immobility
is the way to go. She implies that firm-specific investment and co-specialized assets retard
the mobility of superior resources; managers can thus increase rent appropriation and sustain
their competitive advantage. In contrast, the analysis of Simulation 3 established that the
mobility of resources, in the right direction, is critical to the generation of competitive
advantage. Managers need to think in terms of comparative mobility of resources; as long as
the number and quality of resources entering the firm are more useful than those leaving the
firm, accumulation will ensure that the relevant resource-stocks build up, over time.
This idea also has a parallel with respect to competitors: managers need to ensure that the
quantity and quality of resources entering their firm is better than those being acquired by the
competition to increase their competitive advantage. The hint is that managers could be more
outward oriented and think like entrepreneurs: those continuing to attract the best net quality
of resources will generate higher revenue or lower cost than their competition and would
therefore improve the performance and profitability of their firm relative to their competitors.
These simulations also provide evidence that bringing in the appropriate kind of dynamic
complexity can create value. Earlier in this paper, we looked at some other papers that study
the impact of detail complexity on performance. A comparison of the two modes of value
creation shows up two contrasts. First, detail complexity may protect valuable resources from
being copied because the details are opaque or myriad to competitors while dynamic
complexity relies on the probability that a small difference in initial conditions and/or
interactions can open up large gaps in performance, over time.
Second, due to the linkages between the many details, organizations employing detail
complexity have to compromise on flexibility. Dynamic complexity provides flexibility
because policies can be changed quickly, according to necessity. The implication for senior
management is that in an active environment that prizes flexibility, organizations relying on
dynamic complexity may have a significant advantage over those relying on the other kind.
The facility to use dynamic complexity is thus an example of superior management.
References:
Amit R.H. and P.J.H.Schoemaker (1993). ‘Strategic assets and organizational rent’, Strategic Management
Journal, 14(1), pp. 33-46.
Arthur, W.B. (1989). ‘Competing technologies, increasing retums and lock-in by historical events’, Economic
Journal, 99, pp. 116-131.
Arthur, W.B. (1990). ‘Positive feed-backs in the economy’, Scientific American, pp. 92-99.
Bamey, J.B. (1986). ‘Strategic factor markets: Expectations, luck, and business strategy’, Management Science,
32(10), pp. 1231-1241.
Bamey, J.B. (1989). ‘Asset stocks and sustained competitive advantage: A comment’, Management Science,
35(12), pp. 1511-1513.
Barney, J.B. (1991). ‘Firm resources and sustained competitive advantage’, Journal of Management, 17(1), pp.
99-120.
Bamey, J.B. and E.Zajac (1994). ‘Competitive organizational behavior: Toward an organizationally-based
theory of competitive advantage’, Strategic Management Journal, 15 (Winter Special Issue), pp. 5-9.
Black, J.A. and K.B.Boal (1994). ‘Strategic resources: Traits, configurations and paths to sustainable
competitive advantage’, Strategic Management Journal, 15(1), pp. 131-148.
Collis, D.J. and C.A.Montgomery (1995). ‘Competing on resources: Strategy in the 1990s’, Harvard Business
Review; 73(4), pp. 118-128.
Collis, D.J. (1994). ‘How valuable are organizational capabilities?’, Strategic Management Journal, 15 (Winter
Special Issue), pp. 143-152.
22
Cyert, R. and J.March (1963). A Behavioral Theory of the Firm. Harvard University Press, Prentice-Hall,
Englewood Cliffs, NJ.
DeCarolis, D.M. and D.L.Deeds (1999). ‘The impact of stocks and flows of organizational knowledge on firm
performance:’, Strategic Management Journal, 20(10), pp. 953-968.
Dierickx I. and K.Cool (1989). ‘Asset stock accumulation and the sustainability of competitive advantage’,
Management Science, 35(2), pp. 1504-1511.
Forrester, J.W. (1961). Industrial Dynamics. Pegasus Communications, Waltham, MA.
Grant, R.M. (1991). ‘The resource-based theory of competitive advantage: Implications for strategy
formulation’, California Management Review, Spring 1991, pp. 114-135.
Hamel, G. and C.K.Prahalad (1990). ‘The core competence of the corporation’, Harvard Business Review,
68(3), pp. 79-91.
Levinthal, D.A. (1997). ‘Adaptation on rugged landscapes’, Management Science, 43(7), pp. 934-950.
Levinthal, D.A. and J.Myatt (1994). ‘Co-evolution of capabilities and industry: The evolution of mutual fund
processing’, Strategic Management Journal, 15(1), pp. 45-62.
Lippman, S. A. and R.P.Rumelt (1982). ‘Uncertain imitability: An analysis of interfirm differences in efficiency
under competition’, Bell Journal of Economics, 13, pp. 418-453.
Mahoney, J.T. and J.R.Pandian (1992). ‘The resource-based view within the conversation of strategic
management’, Strategic Management Journal, 13(5), pp. 363-380.
Malerba, F., R.Nelson, L. Orsenigo and S.Winter (1999). ‘History friendly models of evolution:’, International
Journal of Industrial Organization, 19, pp. 635-664.
Malerba, F., R.Nelson, L.Orsenigo and S.Winter (2001). ‘Competition and industrial policies in a history
friendly model of the ....’, Industrial and Corporate Change, 8(1), pp. 3-41.
Mason, R.O. (1969). ‘A dialectical approach to strategic planning’, Management Science, 15(8), pp. B403-
B414.
Miller, D. and J.Shamsie (1996). ‘The resource-based view of the firm in two environments:’, Academy of
Management J ournal, 39(3), pp. 519-543.
Mitroff, I.T., R.O.Mason and V.P.Barabba (1982). ‘Policy as argument - A logic for ill-structured decision
problems’, Management Science, 28(12), pp. 1391-1404.
Morecroft, J.D.W. (1984). ‘Strategy support models’, Strategic Management Journal, 5(3), pp. 215-229.
Morecroft, J.D.W. (1985). ‘Rationality in the analysis of behavioral simulation models’, Management Science,
31(7), pp. 900-916.
Morecroft, J.D.W. (1994). ‘Executive knowledge, models and leaming’. In J.D.W. Morecroft and J.D. Sterman
(eds.) Modeling for Learning Organizations. Productivity Press, Portland, OR, pp. 3-28.
Morecroft, J.D.W. (2000). ‘Resource management under dynamic complexity’. In J.D.W. Morecroft, A. Heene
and R. Sanchez (eds.) A Systems View of Resources, Capabilities and Management Processes.
Nelson, R. and S.Winter (1982). An Evolutionary Theory of Economic Change. Harvard University Press,
Cambridge, MA.
Noda, T. and D J.Collis (2001). ‘The evolution of intraindustry firm heterogeneity: Insights from a process
study’, Academy of Management Journal, 44(4), pp. 897-925.
Penrose, E.T. (1959). The Theory of the Growth of the Firm. Oxford University Press Inc., New Y ork, NY.
Peteraf, M.A. (1993). ‘The comerstones of competitive advantage: A resource-based view’, Strategic
Management Journal, 14(3), pp. 179-191.
Porter, M.E. (1991). ‘Toward a dynamic theory of strategy’, Strategic Management Journal, 12 (Winter Special
Issue), pp. 95-117.
Powell, T.C. (2001). ‘Competitive advantage: Logical and philosophical considerations’, Strategic Management
Journal, 22(9), pp. 875-888.
Priem, R.L. and J.E.Butler (2001). ‘Tautology in the resource-based and the implications of extemally
determined resource value:’, Academy of Management Review, 26(1), pp. 57-66.
Prahalad, C.K. and R.A .Bettis (1986). ‘The dominant logic: A new linkage between diversity and performance’,
Strategic Management Journal, 7(6), pp. 485-501.
Riegel, R. and J.S.Miller (1966). Insurance Principles and Practices. Englewood Cliffs: Prentice-Hall, Upper
Saddle River, NJ.
Rivkin, Jan W. (2000). ‘Imitation of complex strategies’, Management Science, 46(6), pp. 824-864.
Rouse, MJ. and U.S.Daellenbach (1999). ‘Rethinking research methods for the resource-based perspective:’,
Strategic Management Journal, 20(5), pp. 487-494.
Rumelt, R.P. (1984). ‘Toward a strategic theory of the firm’. In R. Lamb (ed.) Competitive Strategic
Management. Prentice-Hall, Englewood Cliffs, NJ, pp.556-570.
Sterman, J.D. (2000). Business Dynamics: Systems Thinking and Modeling for a Complex World.
Irwin/McGraw Hill, New Y ork, NY.
Wemnerfelt, B. (1984). ‘A resource-based view of the firm’, Strategic Management Journal, 5(2), pp. 171-180.
23
APPENDIX 1
Background of the Insurance Industry & Description of its Activities
The beginnings of the modern life insurance industry can be traced back to the 16 century as the
earliest recorded policy was sold in 1583 in Endand. Since then life insurance has become an
international social institution. The aim of the industry is to help cornbine the risks from a large
number of individuals to reduce the degree of uncertainty. These individuals agree to meke srrall
contributions in order to reimburse those who suffer from events insured against. Thus the scope of
the life insurance industry is to enable persons avoid the financial consequences of risks or
uncertain events (Riegel and Miller, 1966).
A typical British insurance firm sells “policies” (insurance contracts) to those who want to insure
themselves. These insurance contracts are the “products of the firm. Policyholders usually pay
relatively small annual “prerriums” and meke claims with the insurance company if and when they
suffer losses covered by their policies. In the meantime, the insurance firm is supposed to invest and
grow the funds that accrue to them in order to keeping them secure, meet pay-out needs and
return funds to mature policy-holders as and when necessary.
Maor players have a common set of tasks - selling insurance contracts, selecting risks, fixing and
collecting premiurrs, writing policies, investing money, keeping accounts, collecting, researching and
analyzing statistics, processing claims and dealing with legal issues and cases. These tasks are
executed either by building the required skills as individual firms, or sharing them from a common
pool - depending upon the quantum of required investment and the scope for differentiation. Given
thet new types of offerings can easily be copied, it is really difficult to establish a sustained
differentiation with respect to competitors. Collecting, researching and analyzing statistics is a
pooled activity, due to econorvies of scale.
The claims department (a cost center) is responsible for processing claims and the legal issues
involved therein. The scope to differentiate here is lirrited!2. The investment department (a profit
center) invests the incoming premiums. Some countries require the financial performance of
investments be legally kept apart from rest of the organization. This brings us to the remaining
activity - the selling of insurance contracts. Demand can be generated through push (sales) and pull
(merketing). But the scope for differentiation through marketing is lirrited, as it is difficult to
compete on prices and very difficult to sell differentiated products (contracts) on a sustained basis.
Thus only a relatively small anount of funds are put aside for marketing campaigns compared to the
funds put aside for the agency department, which handles sales agents. These departments are
shown in Figure 12.
There exist different kinds of agency systems in the insurance industry - e.g the general agency and
branch office systerrs in life insurance, the independent agency and exclusive agency systems in
property and casualty insurance. Common to agency systems is the practice of selling policies
through agents who receive a commission for the sale of these contracts. They differ in the degree
of control that the management of the firm exercises over its agents and their compensation
structure. Agents are responsible for minirnizing ‘lapses’ in the policies they sell. A ‘lapse’ occurs
when the client discontinues payment of premiums towards an insurance contract before the
contract permits.
” No firm would want to establish either a reputation for compromising on payments or take a hit on its
profitability by relaxing payment standards.
24
FIGURE 12 - An Insurance Firm
Reject or
Accept
ACTUARIAL CLAIMS INVESTMENT
Dept. Dept. Dept.
Pricing
of Investment
Policies Claims of Cash
[Basket of
| Policy ;
‘Variations . |
Prémiums | Cash _
| Inflows
‘Sales
New H
kinds of |
Policies :
MARKETING
Dept.
APPENDIX 2
Measures of Profitability in the Insurance Industry
Insurance industries have their own particular measures of profitability. The measure
of new business profitability is an efficiency measure called New Business Expense
Ratio (NBER) and is defined as:
NBER = Expenses Incurred for Obtaining New Business x 100
New Business Annual Premiums
Another measure of profitability concerned with the efficiency of renewing policies is
called the Renewal Expense Ratio (RER). It is defined as:
RER =(Total Expenses - Expenses Incurred for Obtaining N ew Business)
(Tota Premiums - New Business Prerniurs) / 100
25
APPENDIX 3
Activities in a Typical Agency Department
We focus on the agency department of large firms in the insurance industry and will henceforth
refer to the stylized department as a firm. The firm competes to sell policies to those who want to
buy insurance. These products provide the insurance firm with premiums for the length of the life of
the product, if they do not ‘lapse’. The larger the product base (i.e. the inventory of live policies sold
by the firm), the larger the cash flow and revenue to the insurance firm These sales are actually in a
cycle of three stages. In the first stage, agents are recruited from the market to be employees of the
firm and become part of the agent body that goes out to sell policies to prospective customers in
the market. Firms always seek to hire more experienced agents from the market and to retain the
better-performing agents. In the second stage, policy sales accumulate as the body of policies in
force and form the basis of the future revenue stream (as prerriums) unless these policies lapse.
Policy sales and lapses are a function of the skill level of the agents. In the third stage, agents are
compensated based on the sales made and lapses occurred in that particular year.
When agents join the firm, they not only increase the headcount of sales employees but also add
their sales skills to the skill pool of the firm Also, from time to time, some agents quit the firm and
some are promoted. These decrease the headcount of the sales employees as well as from the
aggrecete skill pool of the firm While agents quitting the firm tend to have lower than the average
skill level, those promoted will have a higher than average skill level. Thus it is a challenge to the
management to maintain and improve upon the skill level of their agent base. Agent compensation is
of particular inmportance. If the agents perform above the performance level expected by the market,
the compensation that results is also above market expectations. Conversely, performance inferior
to market expectations leads to inferior compensation. In turn, compensation affects the quit rate of
agents (the quit rate influences the lapse rate of new policies sold) as well as the attractiveness of
the firm to new agents who are considering whether to join the firm
We use a measure of performance that reflects the costs and benefits of net new business acquired
and which is closely related to the measure of new business expense ratio described in Appendix 2.
It is defined as the ratio of net new products to the expenses incurred in selling these policies. We
ignore products that mature because they are a function of the firm's existing portfolio of products
and therefore, an indication of the past performance of the sales force. Hence,
Profitability = Number of products sold - Number of products lapsed
Total expenses of selling products
As our focus is on the agency department, we use the quantum of new products sold per year in lieu
of the premiums earned annually due to new business. As we assume products of standard length and
premium, the only dimension of distinction in sales is the number of products sold. Hence, larger the
number of contracts the firm can sell (that do not lapse), per unit labor cost, the greater the
profitability of the firm
APPENDIX 4
Headcount
Figure 13 shows the processes directly concerned with the management of agent headcount, using
system dynamics diagran symbols. The number of agents in the firm (Agents) is a tangble resource-
stock that is affected by the flows of agents who quit, who are promoted and who are hired. This
flow of agents who quit (Agent Quits) is a function of the (Standard) Agent Quit Rate and is also directly
proportional to the number of agents in the firm. Agents may quit because of a variety of reasons:
boredom, better opportunities, unsatisfactory compensation / unsatisfactory performance, etc.
These factors have a clear innpact - changes in the factors can increase or decrease Agent Quit Rate.
We'll explore these factors in more detail later. Sirrilarly, the resource-stock Agents also decrease
26
because of the flow of promotions granted by management (Agents Promoted). Agents are promoted
to managers. Managers have other responsibilities besides selling policies but at this moment we will
assume that managers are external to the system. The number of managers required (and hence the
rate of agents promoted) will be proportional to the nurmber of agents thet exist in the firm We
also assume thet a fixed proportion of agents is promoted (Prornation Rate).
Agents lost due to quits and promotions have to be replenished. The process of hiring does this.
The absence of hiring would result in a continuous decrease in the number of agents available for
sales. Hence the rate of agents to be recruited (Hires) is determined by the number of agents who
quit and who are promoted. As long as this equation holds, there is no net increase or decrease in
the number of agents available to the firm and steady state conditions prevail. It is relatively easy to
maintain headcount because of the ease of counting persons and then set a target to hire the
required number. The number of agents quitting the firm, the number of agents being promoted as
well as the number of agents itself have an important impact in accounting for the dynarics of the
skill subsystem of the firm, which we explore in detail next.
Skill Subsystem
This subsystem models the dynamics of the agents’ skills. Figure 14 shows the processes directly
concerned with the management of agent skills, which is an intangble resource for the firm. All the
agents in the firm have some intrinsic sales skill (measured in terms of equivalent years of
experience) and the aggregate skill amnong the agents is accounted by the stock Agents’ Sales Skill (the
skill pool of the firm). From this skill pool and the number of agents, we can work out the average
skill level of the firnn’s agents (Skill per Agent). This measure of ability has an important impact on
sales and other functions of the firm
When agents quit the firm, they carry away their skill. The skill lost (Lost Skill from Quits) depletes the
skill pool of the firm Apart from being a function of Agent Quits, Lost Skill from Quits depends also on
the Relative Skill of Quits. The ideal objective of managers is to let this relative skill reflect the average
skill level of the market, provided they can retain their superstars. In reality, however, it is likely to
be managed as a simple function or proportion of the existing skill in the firm (Skill per Agent). If so,
this number would be less than one as we expect the agents performing worse than the firm average
to be prompted to leave (known in industry parlance as ‘culling). Keeping this policy simple benefits
both management and the agent-employees. For management, it reduces some uncertainty about the
amount of skills they expect to lose and reduces the extra effort implied in implementing a policy
that varies in accordance with other phenomena in the firm A simple policy provides greater
transparency for agents as well as job-security for those agents performing above average.
When agents are promoted, the skill pool is sirrilarly depleted by Lost Skill from Promation. As
expected, Lost Skill from Promotion depends on Agents Promoted and the Relative Skill of Promoted
Agents, which we assume to be a simple ratio: in reality it is proportional to the existing skill in the
firm or to be a constant. If the ratio is in terms of Skill per Agent, it would be greater than one as we
expect only those agents who are better than average to be promoted. Once acain, a simple policy
here benefits both management and agents in the manner outlined immediately above.
Agent skills lost due to quits and promotions also have to be replenished. This is done from through
the skills of those hired (Added Skill at Hire). The absence of hiring would result in a continuous
decrease in the skill base of the firm The added skill is a function not only of the number of agents
hired but also of the average skill level of the agents hired (Agent Skill at Hire). This skill is determined
by two influences. The first is Standard Agent Skill at Hire, which basically reflects the market
conditions about the skill of agents available for hire to all firms in the industry. Rarely can one
expect this market average to be above the average merket skill level - as above average agents are
easily hired away by other firms too. The second influence is the effect of compensation on
recruited skill (for the firm in question). The overall effect of this is to move the actual average skill
27
FIGURE 13 - Headcount Subsystem
hires
agents «——— promotion
promoted p rate
agent
auyts
agent quit
rate
COMPENSATION SUBSYSTEM
FIGURE 14 - Skill Subsystem
ren
—_ _ SUBSYSTEM _)
SUBSYSTEM agents quits skill per
\ " /e- agent
\ SUBSYSTEM
\ hires oo
\ relative skill, lost skill
of quits from quits agents
0) Qe Agents! 4" promoted
ss added skill [Sales Skil —2—?3
a ve lost skill from
1 promotions
ele tt level of
agent skill» '©} lative skill of
oat hire nie agents promoted agents
COMPENSATION. |.---impact of recent
SUBSYSTEM compensation
28
FIGURE 15 - Productivity Subsystem
standard sales agent s_| HEADCOUNT
- coset | SUBSYSTEM |
aA Products
sales. _ - produ f FOIE: a cts 8
productivity et on nd
COMPENSATION |... pl ene }
SUBSYSTEM finete
relative maturity
agent skill
4 lapse
a ratew_ ,
/ Skill per ss impact of
i agent “agent quit rate
SUBSYSTEM SUBSYSTEM
FIGURE 16 - Compensation Subsystem
time to react to
SKILL compensation
SUBSYSTEM |
coca recent relative impact of
impact of compensation recent comp
recent comp v oo
‘ 7
expected agent relative
— : standard i
compensation momen agent ak pagent cut HEADCOUNT
| SUBSYSTEM
agent fixed__, compensation
compensation per agent
relative agent = max agent
t quit rate—— quit rate
case size——® variable agent /
compensation /Ampact of
A *-.. lapse rate /agent quits proportion of
\ ue agent quit
agent aan ; -“ policies lapsing
eonmnetssl2n) eroductivity PRODUCTIVITY woe
rae SUBSYSTEM Js 29
of the agent hired up and down in accordance with the average relative compensation (relative to
market expectations). We will explore this in detail in a later subsystem.
Taken together, these inflows and outflows of skills represent a significant challenge to management
trying to preserve and grow the skill base of the firm It is difficult to measure something more
intangble like the skill associated with each agent who leaves, and more so to set a target based on
those who enter the skill pool - since it is tricky to recognize and induce only the most talented to
join the firm. Since the rate at which agents quit and the subsequent rates at which agents have to be
hired are likely to fluctuate and vary with time, management would find it more practical to manage
this balance with simple and robust policies that reduce their effort and provide greater
transparency to all.
Productivity Subsystem
This subsystem elaborates on the productivity of the sales agents and the resultant dynamic impact
on the product portfolio. Figure 15 sketches the relevant process details. Skill per Agent from the
previous subsystem determines Relative Agent Skill which is basically a ratio of Skill per Agent to the
prevailing expected average market standard skill of sales agents, Standard Agent Skill. In turn, Relative
Agent Skill determines Sales Productivity, which is a product of Relative Agent Skill and Standard Sales
Productivity, the productivity of the average sales agent in the industry in terms of policies sold per
year. From here, Product Sales is simply the product of Sales Productivity and Agents, from the
headcount subsystem and indicates the number of policies added (in a given amount of time)to the
existing portfolio of policies, Products in Force. More sales imply an increase in the number of Products
in Force, which is a tangble resource of the firm. The number of Products in Force may also decrease,
because of two different effects. The first effect is a sort of expected, natural decay as policies
meture (Products Maturing) and prerniums for those policies stop coring in. The fraction of Products
in Force that matures is inversely proportional to Time to Maturity, which we assume to be a constant
in our simplified model. The second effect that decreases Products in Force is due to Products Lapsing
Since most lapses are likely to occur in the very early life of the policy, it can be estimated to be
proportional to Product Sales (not Products in Force) and the Lapse Rate.
As mentioned earlier, a policy laoses when the customer stops paying the requisite premium for the
policy on time. This phenomenon may be exacerbated if the customer discovers that he has been
sold a policy that is not the best for his needs. In other words, it is a function of the skill of the agent
to listen to the customer and draw out his needs - a skill that can be estimated by the experience
that an agent has. In the model, Lapse Rate can initially be assumed to be a function of Maximum
Lapse Rate (a constant) and Relative Agent Skill. Simply put, an increase in the Relative Agent Skill means
that the agents are more skilled than market expectations and the Lapse Rate drops as a result.
Conversely, a decrease in the Relative Agent Skill implies that the agents are less skilled than market
expectations and the Lapse Rate increases. In all these three flows that affect the accumulation of
Products in Force there are really very few direct levers for management to influence.
Compensation Subsystem
This subsystem models the mechanics of compensation for agents and outlines its impact on the
other three subsystems. Figure 16 shows how compensation is influenced by Sales Productivity and
Lapse Rate from the productivity subsystem, and through what variables it can influence the other
three subsystems. Meanwhile the important variables in this subsystem are Variable Agent
Corpersation, Relative Agent Conpensation, Recent Relative Corpensation and Agent Quit Rate. Before
we go on to understand the intricacies of agent compensation, it is useful to rerrind readers that in
our simplified mode! the only aspect through which a sales agent can distinguish himself is by making
more sds, as we have assumed uniform products (insurance policies). Uniform policy length and
uniform premiums follow from this assumption. The total compensation per agent (Total Agent
jon) is the sum of two types of compensation - variable (Variable Agent Compensation) and
fixed (Fixed Agent Compensation). Variable Agent Compensation is determined by four (multiplicative)
factors. The first factor is obviously the average productivity of the agents, Sales Productivity. The
30
second and the third are the size of the premium (Case Size) and the Agent Comission Rate. While
Case Size is assumed to be a constant, managers would prefer to follow the industry standard of
paying the agent 100% of the commission collected from a policy in the first year of the sale (if it
does not lapse). The last factor is the complement of the Lapse Rate (i.e. 1.0 - Lapse Rate), as the
agent will be paid commission only for the policies that do not lapse. Fixed Agent Cornpensation is a
relatively small component of the total expected compensation (about 259, paid by management at
the level of the industry standard and does not need to change out of step with the industry.
Given whet other agents are earning in other firms on average (Expected Agent Conrpensation), a
relative compensation (Relative Agent Cornpensation) can be calculated that is the ratio of Total Agent
Compensation to Expected Agent Compensation. One might expect that agents decide on whether they
want to stay on, quit or join based on this variable, but in reality, agents prefer to wait and watch a
little, partly in the hope that things may change for the better and partly to be sure that they are
indeed getting a good/bad deal from the firm. In practice, it requires about three months to
crystallize expectations. Effectively this implies that the agents tend to act on the compensation
status three months previous to the current date. The actual variable that motivates the action of
the agents is called Recent Relative Compensation.
Let us now examine in detail how Recent Relative Compensation impacts the variables in the other
subsystems, viz. Agent Skill at Hire and Agent Quit Rate - to get an idea of the overall impact to the
system brought about by the dynarrics of this key variable. Recall that the second component of
Agent Skill at Hire was the effect of cormpensation on recruited skill. When Recent Relative
Compensation is equal to 1, (i.e. average compensation is as expected by the industry benchmark)
agents perceive themselves to be no better off by joining it has no effect on the variable in question.
Agents that interview with the firm are indifferent towards the firm, as they perceive it to be neither
better nor worse off than competition. When Recent Relative Conpensation is greater than 1, implying
superior average compensation, the overall effect is one of positive reinforcement. Agents being
recruited will now expect to earn above average and the better agents will try to beat out those
with lower skills and attermpt to enter the firm. This, in turn, innplies an increased value of Agent Skill
at Hire because it is a product of Standard Agent Skill at Hire and the effect, which is now positive (i.e.
> 1). In the same manner, when Recent Relative Corrpersation falls below 1, the better agents shun
the firm because of lower expectations. The reinforcement effect turns negetive when its value falls
below 1 and has a deleterious effect on Agent Skill at Hire.
Just as Added Skill at Hire is composed of two factors - Standard Agent Skill at Hire and the effect of
compensation on Agent Skill at Hire, Agent Quit Rate may also be seen to share the same structure. In
this case it would be the product of the invariant Standard Agent Quit Rate and the effect of
compensation on the quit rate. Recent Relative Compensation has a similar impact on Agent Quit Rate,
but in a nominally opposite direction. When it climbs beyond 1, agents in the firm want to enjoy the
good times (i.e. cormpensation above industry average) and are reluctant to leave the firm- asa
result of which the effect of compensation on the quit rate falls below 1, and with it falls Agent Quit
Rate. Similarly, when Recent Relative Compensation falls below 1, agents are keen to seek out
opportunities elsewhere: consequently Agent Quit Rate rises above the value of Standard Agent Quit
Rate. There is also the cascading effect of Agent Quit Rate on Lapse Rate. An increased Agent Quit Rate
implies higher agent turnover, and in many cases, agents that leave the firm may convince their
customer to pay the premium to their new employer instead of the old one. As a result the Lapse
Rate for the old employer increases. Similarly, a drop in the Agent Quit Rate implies a drop in the
Lapse Rate as well. We have seen earlier that all these three variables - Agent Skill at Hire, Agent Quit
Rate and Lapse Rate impact different parts of the system in turn and the overall dynavic impact of a
shift in Recent Relative Corpensation away from the stable value of 1 is not so easy to predict.
However, it is pointed out that as in the previous subsystem, there are again very few levers for
to play around with on along-term or medium-term basis. For example, reducing the
fixed salary or comission will bring about unfavorable comparisons with the rest of the industry
while increasing these commissions for all agents are hard to justify because of increased expenses.
31
Profitability: The calculation of Profitability is relatively simple. It is simply the difference between
Product Sales and Products Lapsing, divided by Total Agent Conpensation.
APPENDIX 5
Exploring the Dynamics of the Compensation Sub-system
Next we explore the dynarrics of the compensation subsystem by itself (Figure 18), which includes
one positive and one negative feedback loop. Unlike the collection of sub-systems, which is modeled
for Simulation 1, relevant policies are usually not prescribed for this sub-system. Thus, the objective
of studying the behavior is to determine whether Level of Hired Agents can be a potential source of
differential performance and understand its practical implications. The names of the variables in Fig.
18 appear to differ from those in Figure 16 because the emphasis here is more on the loop and less
on the intermediate variables. Fixed compensation is set at $5000 per year and expected total
compensation is $20000 per year. The annual premium from the standard policy (Case Size) is set at
$160. To understand the behavior of the subsystem, we will analyze the outcome of the subsystem
(Recent Relative Compensation) across a range of values for the input (Relative Skill per Agent) - 0.8, 0.9,
1.0, 1.1, 1.2. Figure 17 shows the dynarric behavior of the isolated subsystem over ten years. We
observe that when the input is different from 1, there is an initial tendency of divergence followed by
a stable steady state. We infer that for a given difference of the incorring value from 1, the value of
the output stabilizes itself at a value greater than the initial difference from 1.
We explain what is happening in this subsystem by tracing the path of the loops starting froma
convenient place - in this case Relative Corrpensation. Irrespective of the value of the input, the initial
value of Recent Relative Corrpensation is 1. When the input (Relative Compensation) is different from 1,
Recent Relative Cornpensation shifts towards the value of Recent Cornpensation. If it is greater than 1, it
lowers the value of Agent Quit Rate which then lowers the value of Lapse Rate below its standard
vdue. This new lower value combines with Sales Productivity to increase Variable Agent Compensation
to a slightly higher value than its previous value. Consequently, Relative Compensation increases
beyond its original value. This continues in an iterative feshion until the difference between Relative
jon and Recent Relative Compensation shrinks away. It is useful to note that this gap is not
obliterated in the first iteration itself. This accounts for the delayed, yet restrained pattern of
divergence. W hen Relative Skill per Agent is less than 1, a symmetric process of adjustment takes
place in the reverse direction until Recent Relative Compensation reduces and reaches the value of
Relative Corrpensation at something less than 1.
The find key consequence of Recent Relative Compensation attaining a value different from 1 is the
multiplier (non-linear) effect of Recent Relative Compensation on Leva of Hired Agents. This effect
amplifies further the aready seen deviation effect (i.e. the deviation effect shown by Recent Relative
Corpersation as a function of the deviation from Relative Skill per Agent, described in the previous
paragraphs). The graph in Figure 19 gives an indication as to the magnitude of deviation suffered by
Level of Hired Agents due to deviations in Recent Relative Compensation. N ote that all the relationships
between the different variables in this sub-system have been subject to empirical evidence.
To summarize, when there is a deviation from the steady-state input value of Relative Skill per Agent
to this sub-system, there is an increased deviation in the output, Level of Hired Agents. This increase
in deviation can, under appropriate contexts, be the source of a competitive advantage to firms. The
practical implication is that if a firm has an average skill level that is slightly higher than the average
skill level expected by the rrarket, then it finds it easier and easier to attract agents of even higher
skills. Conversely, if a firm has an average skill level that is slightly lower than the average skill level
expected by the market, then it finds it more and more difficult to hold onto its skilled agents, who
are replaced by new agents with inferior skills.
32
FIGURE 17 - Compensation Characteristics
Graph for the effect of recent relative
compensation on recruited skill
1.40).
1.26)
1.12
0.98
0.84 |*
0.70
0 10
Initial Value of Relative Agent Skill: [WR eee
Initial Value of Relative Agent Ski: 1.1 == ==============e8=8
Initial Value of Relative Agent Skill: 1.0
Initial Value of Relative Agent Skil: 0.9 —— — —= = — — — — — — =
Initial Value of Relative Agent Skill: 0.8 = = = ee ee ee
Units: equivelent years of experience (above) and dirrensionless (below)
Graph for agent quit rate
0.26
(24 en
0.22 |=
0.20
0.16
0 2.5 5 7.5 10
Time (year)
33
FIGURE 18 - Compensation Sub-system
_ relative—
standard fr agent skill
agent skill
skill ra ivi
per agent
+
2 lapse
variable ratewt
fixed ompensation
compensation me
Nhe i
tol itrate
standard a
agent skill at vert
hire
— ra iol
recen
level of compensation
+ agen
agents rs
FIGURE 19 - Effect of Recent
Compensation on Level of Hired Agents
12
Relative Recruited Skill
Level is an intermediate
step and an antecedent of
Level of Hired Agents
Relative Recruited
uv
Skill Level .
4 8
Recent Relative Compensation
Back to the Top 34
PROGRAM LISTING
relative agent skill=
skill per agent/ standard agent skill
~ 1
= |
Net Products In Force= INTEG (
net product sales - products maturing, initial products)
~ products
average product life=
Net Products In Force / (products maturing)
~ year
= |
Profitability=
net product sales / total compensation * 100*case size
~ |
net product sales=
product sales * (1-lapse rate)
~ products/year
= |
product sales=
sales productivity * A gents
~ |
lapse rate=
effect of skill per agent on lapse rate * effect of agent quits on lapse rate* adjuster
1
= |
Agents’ Sales Skill=INTEG (
added skill at hire - lost skill from agent quits - lost skill from promotion + agent
leaming, A gents*agent initial average skill level)
ms equiv years
~ Agent Sales Skill is an intangible resource that determines how effective each
sales agent is at finding and converting sales prospects into closed deals
|
hires=
target number of recruits
~ agents/year
~ The rate at which agents are taken into the firm - measured now in agents per
|
agents promoted=
0.024*A gents
~ agents/year
~ |
expected agent compensation=
20000
~ $/(year*agents)
year.
sales productivity=
standard sales productivity *relative agent skill
~ products/(year* agents)
target number of recruits=
(agent quits + agents promoted)
~ agents/year
~ |
total compensation=
total agent compensation
~ $/year
effect of skill per agent on lapse rate=
tel agent skill lookup(relative agent skill)
~ Fraction
~ |
adjuster=
2.5
~. ; |
rel agent skill lookup(
[(0,0)-(5,1)],(0,1),(1,0.6667),(2,0.45),(3,0.25),(4,0.15),(5,0.1))
agent skill at hire=
standard agent skill at hire * effect of recent relative compensation on recruited skill
~ equiv years/agents
added skill at hire=
hires * agent skill at hire
~ equiv years/year
effect of recent relative compensation on recruited skill=
effect of rel comp on recruit skill lookup(recent relative compensation)
~ Dimensionless
effect of rel comp on recruit skill lookup(
[(0,0)-(8,12)],(0,0),(0.11,0.03),(0.22,0.07),(0.31,0.17),(0.58,0.5),(1,1),(1.1,1.21)\
(1.5,1.8),(2.5,3.25),(5,7),(8,12))
~ Dimensionless
= |
standard agent skill at hire=
3
~ equiv years/agents
PI init= INITIAL(
Profitability)
Tot Comp init= INITIAL(
total compensation)
oat |
total agent compensation=
Agents * compensation per agent
= $/year
ms |
Profit Index 1=
net product sales / total compensation * 100
~ (products*year)/$
initial average product life=
6
~ year
a |
initial products=
initial average product life * net product sales
~ products
~ |
unitless agent quit rate=
agent quit rate / max agent quit rate
~ Fraction
a |
max agent quit rate=
1
~ l/year
effect of agent quits on lapse rate=
unitless agent quit rate * proportion of agent quit policies lapsing * impact of agent
relationships on lapse rate
~ Dimensionless
~ (1* (1- unitless agent quit rate) + unitless agent quit rate * \
proportion of agent quit policies lapsing * impact of agent relationships \
on lapse rate ) - Maurice’s formulation- only 1st part OK
|
recent relative compensation=
DELAY 1I (relative compensation, time to react to compensation , 1)
~ 1.
ot |
time to react to compensation=
0.25
~ year
= |
relative skills of promoted agents=
1.8332
~ 1.
effect of relative comp on agent quits=
effect of relative comp on agent quits lookup(recent relative compensation)
~ 1
effect of relative comp on agent quits lookup(
[(0,0)-(4,2)],(0,2),(0.5,1.5),(1,1),(2,0.5),(3,0.25),(4,0.1))
~ 1
~ |
compensation per agent=
variable agent compensation +agent fixed compensation
~ $/(year*agents)
= |
agent quit rate=
standard agent quit rate *effect of relative comp on agent quits
~ l/year
lost skill from promotion=
agents promoted * skill per agent *relative skills of promoted agents
~ equiv years/year
= |
variable agent compensation=
agent commission rate * case size * sales productivity * (1 - lapse rate )
~ $/(year*agents)
~ The average yearly compensation of the agent is a key variable in \
determining the rate at which agents quit the firm.
|
telative compensation=
compensation per agent / expected agent compensation
~ Fraction
relative skill of quits=
0.9
- Fraction
os I
lost skill from agent quits=
agent quits * skill per agent * relative skill of quits
~ equiv years/year
impact of agent relationships on lapse rate=
- Fraction
os |
agent commission rate=
1
~ Dimensionless
2 |
agent quits=
Agents * agent quit rate
~ agents/year
~ The rate at which agents leave the firm is measured in agents / year and \
is very important for it to be at a low rate to maintain the skill level \
of the firm.
case size=
160
~ $/products
= I
standard sales productivity=
125
~ products/(year* agents)
proportion of agent quit policies lapsing=
0.75
~ Fraction
= |
standard agent skill=
3
~ equiv years/agents
~ |
standard agent quit rate=
0.2
1/year
a |
products maturing=
Net Products In Force / time to maturity
~ products/year
~ |
time to maturity=
20
~ year
i |
manager person convertor=
1
~ managers / person
Agents= INTEG (
hires - agent quits - agents promoted,
100)
~ agents
~ Agents are a critical resource because they are the direct generator of
revenues. The agent job invovles visiting clients with the hope that they will persuade the \
client to purchase a financial service product. A typical branch would \
have 1 branch manager with 5-10 sales managers working for him/her. Each \
sales manager would be responsible for 6 - 12 agents.
Only potential sustainable advantage from complementarity involving \
skills, leaning ablitility, company reputation, manager time
agent initial average skill level=
3
~ equiv years/agents
agent fixed compensation=
5000
- $/(year*agents)
~ Fixed component of compensation e.g. salary
|
skill per agent=
Agents’ Sales Skill / Agents
~ equiv years/agents
~ The skill per agent is derived by dividing the total amount of experience \
by the number of agents to get units of equiv_years per agents.
FES GEASS SAFARIS SIO S EI IS AISA {CC 111112 K KKK 2K 2K 2K 2K 242K
.Control
SCC
Simulation Control Paramaters
FINAL TIME =20
~ year
~ The final time for the simulation.
|
INITIAL TIME =0
~ year
~ The initial time for the simulation.
|
SAVEPER =0.125
~ year
~ The frequency with which output is stored.
|
TIME STEP =0.03125
~ year
~ The time step for the simulation.
