System Dynamics Review

System Dynamics Review vol 27, No 3 (July–September 2011): 219–243

Published online 14 July 2011 in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/sdr.462

Reflections on the foundations of system dynamics

George P. Richardson*

Abstract
Jay W. Forrester’s original statement of the foundations of system dynamics emphasized four ‘threads’: computing
technology, computer simulation, strategic decision making, and the role of feedback in complex systems. Subsequent
work has expanded on these to expose the significance in the system dynamics approach of dynamic thinking, stock‐
and‐flow thinking, operational thinking, and so on. But the foundation of systems thinking and system dynamics lies
deeper than these and is often implicit or even ignored: it is the “endogenous point of view”. The paper begins with
historical background, clarifies the endogenous point of view, illustrates with examples, and argues that the
endogenous point of view is the sine qua non of systems approaches. What expert systems teachers and practitioners
have to offer their students and the world is a set of tools, habits of thought, and skills enabling the discovery and
understanding of endogenous sources of complex system behavior. Copyright © 2011 System Dynamics Society.

Syst. Dyn. Rev. 27, 219–243 (2011)

The fault, dear Brutus, is not in our stars, but in ourselves.

Cassius, in Shakespeare’s Julius Caesar (1599)

Man is not the creature of circumstances. Circumstances are the creatures of men.
Benjamin Disraeli, Vivian Grey (1827)

We have met the enemy and he is us.

Walt Kelly’s Pogo, on the first Earth Day poster (1970)

Introduction and purpose

There are many published definitions or characterizations of the field of system dynamics
—a search of the Internet reveals a confusing variety—and many more capsule
descriptions embedded in introductions and methods sections of journal articles,
conference papers, and dissertations. One can even find published sources that assert
there is no precise definition. Among the various characterizations, one is sure to find
phrases about stocks and flows and feedback loops, something about dynamics or change
over time, sometimes some references to modeling environments, but the essential
perspective that lies at the core of the field is almost never revealed. It is the purpose of
these “reflections on the foundations of system dynamics” to bring that essential core
perspective vividly to light, with the goal of enhancing practice in the field itself and

* Correspondence to: George P. Richardson, Rockefeller College of Public Affairs and Policy, University at Albany, State
University of New York, Albany, NY 12222, U.S.A. E-mail: [email protected]
Received 17 January 2011; Accepted 2 March 2011

Copyright © 2011 System Dynamics Society

220 System Dynamics Review

clarifying for others the important contributions it can make to policy and theory building.
The reflections begin appropriately with history, and with Forrester’s own early
statements on the nature of the field.

The four foundations

In his seminal article in the Harvard Business Review (Forrester, 1958), Forrester laid out
his initial statement of the approach that would become known as system dynamics. He
founded the approach on what were then four recent developments: advances in
computing technology, growing experience with computer simulation, improved under-
standing of strategic decision making, and developments in the understanding of the role
of feedback in complex systems.
He phrased the four foundations slightly differently in Industrial Dynamics, but the
emphasis was much the same (Forrester, 1961, p. 14):

• the theory of information feedback systems;

• a knowledge of decision‐making processes;
• the experimental model approach to complex systems;
• the digital computer as a means to simulate realistic mathematical models.

The slight change in the list of four here shows Forrester emphasizing an experimental
approach to understanding the dynamics of social organizations that would presumably
be enabled by iterative computer simulation.
Thus we have the familiar and enduring cornerstones of the system dynamics approach
as they were expressed at the founding of the field. But within 10 years Forrester
expressed the foundation quite differently.

The foundation 10 years later

In 1968 Forrester published his classic paper “Market growth as influenced by capital
investment,” now known throughout the field simply as the “market growth model”
(Forrester, 1968a). During that time he was working with former mayor of Boston John
Collins and others on urban problems, resulting in his first non‐corporate study, Urban
Dynamics (Forrester, 1969) and on his teaching text Principles of Systems (Forrester,
1968b). In all of these he began by presenting the structure of the approach he used, not as
the four threads outlined above but rather as a four‐tiered structural hierarchy:

• Closed boundary around the system

‐ Feedback loops as the basic structural elements within the boundary
• Level (state) variables representing accumulations within the feedback loops
• Rate (flow) variables representing activity within the feedback loops
‐ Goal
‐ Observed condition

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 221

‐ Detection of discrepancy
‐ Action based on discrepancy

(Forrester, 1968a, p. 83, 1969, p.12; see also Forrester, 1968b, pp. 4‐17).
To current practitioners, the hierarchy looks very familiar—so familiar in fact that it is
easy to miss its significance. It emphasizes feedback loops, of course, and the familiar
levels (stocks) and rates (flows), all of which are the stock‐in‐trade of all system dynamics
practitioners. Within rate variables it exposes the classic goal‐seeking feedback structure
that exists in some form in all purposeful decisions people make.
However, in this hierarchy it is easy to miss the item that appears at the top of the list:
the closed boundary around the system. But that phrase signals what may be the most
significant part of the system dynamics approach for understanding complex systems. It
signals Forrester’s endogenous point of view.1

The endogenous point of view

Forrester initially phrased it this way:

Formulating a model of a system should start from the question “Where is the
boundary, that encompasses the smallest number of components, within which the
dynamic behavior under study is generated?” (Forrester, 1968b, p. 4‐2).

The idea became summarized in his first principle of system structure, the principle of
the closed boundary:

In concept a feedback system is a closed system.2 Its dynamic behavior arises within its
internal structure. Any action which is essential to the behavior of the mode being
investigated must be included inside the system boundary. (Forrester, 1968b, pp. 4‐1, 2).

Consider the reach of these statements. They tell us to build models that are capable of
deriving the dynamic behavior of interest solely from variables and interactions within
some appropriately chosen system boundary. They tell us not to depend upon any
exogenous forces to produce the dynamics of interest. Moreover, they suggest guidelines
for thinking: they tell us to try to think of dynamics that way, to try to understand system
dynamics as generated from within some conceptual, mental boundary.
In Urban Dynamics Forrester apparently felt the idea was sufficiently important to
warrant a graphic of its own (Figure 1). He explained, “The closed‐boundary concept
implies that the system behavior of interest is not imposed from the outside but created
within the boundary” (Forrester, 1969, p. 12).

Endogeneity and feedback

The most salient aspects of the system dynamics approach are undoubtedly stocks and
flows and feedback loops. These visible elements stand out and command our attention.
But it is worth noting that feedback loops are really a consequence of the endogenous
point of view.

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
222 System Dynamics Review

Fig. 1. Figure reproduced from Forrester (1969, p. 13) symbolizing the closed‐boundary concept

Figure 2 illustrates the idea. On the left is a picture of some simple causal system, with
causal elements tracing ultimately outside the system boundary. The dynamics of variables
A–E are generated partly by interactions among them inside the system boundary but really
stem mainly from variables P, Q, R, and S outside the boundary. The dynamics of this
system are generated exogenously by forces outside the system boundary.
On the right of Figure 2 is an endogenous view, in which the dynamics of variables A–E
are generated solely from interactions among those variables themselves, within the
system boundary. We note from the figure that taking an endogenous point of view forces
causal influences to form loops. Without loops, all causal influences would trace to
dynamic forces outside the system boundary.3 Feedback loops thus enable the
endogenous point of view and give it structure (Richardson, 1991, p. 298).

Endogeneity in practice

Two examples from Forrester’s early writings serve to show vividly what the endogenous
point of view looks like in practice and why it is so crucial in thinking and modeling for
policy analysis. Although the examples are very familiar to system dynamics

Fig. 2. Left: exogenous view of system structure; causality traces to external influences outside the system boundary.
Right: endogenous view; causality remains within the system boundary; causal loops (feedback) must result

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 223

practitioners, they deserve our attention here as we try to go deeply into the nature and
significance of the endogenous point of view.

Market growth as influenced by capital investment

The first example, shown in Figure 3, is the famous market growth model mentioned above
(Forrester, 1968a). The figure vividly shows Forrester’s endogenous point of view: Figure 3(a)
shows a structure with no dynamic influences coming from outside the system boundary. The

Fig. 3. (a) The structure of Forrester’s market growth model (Forrester, 1968a). Note no external dynamic
influences. (b) Behavior generated endogenously by the market growth model, showing a firm in an unlimited
market declining from month 50 on because of its own flawed internal capacity ordering policies

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
224 System Dynamics Review

dynamics shown in the graph in Figure 3(b) stem solely from interactions among the variables
shown and the balancing and reinforcing feedback loops they form together.
Forrester developed this model to illustrate a potential source of corporate poor
performance: the internal operating policies of the company itself. To make the story most
vivid, he set the company represented by the model in a potentially infinite market: there
is no external market cap in the model that might limit corporate growth, production
capacity, the size of the salesforce, or the amount of orders booked per month. Yet the
dynamics shown in Figure 3(b) show a company in which production capacity is
declining, orders booked have a pronounced oscillatory and eventually declining pattern,
and the size of the sales force appears to peak and decline, even though there is no
external limit to the size of the market the salesmen might reach. In an unlimited market,
the company is going out of business.
The key to these dynamics is the policy the company is using to determine when to add
production capacity. The policy (captured in the “Delivery delay condition” at the right of
Figure 3(a)) says to add capacity when the delivery delay for the company’s product
exceeds a target, the “Delivery delay operating goal”. If that goal is based on past
performance, as is the case in this simulation, then it tends to slide upward as the actual
delivery delay rises. As orders booked rise, the delivery delay initially rises, putting
pressure to expand capacity, but that pressure depends on a comparison of the delivery
delay recognized by the company and its target delivery delay. Since that target is based
on past performance, it too is rising slightly, relieving somewhat the pressure to expand
capacity. In this simulation, the sliding delivery delay target never generates enough
pressure in the delivery delay condition to expand capacity, and the company proceeds to
lose market share. It is likely that corporate decision makers in such a setting would
believe the causes of the declining sales trace to an overall declining market, when in fact
(in this model) that market is potentially infinite.
There are three key insights here that should inform our understanding of the
endogenous point of view:

• First, we see in Figure 3(a) an unmistakable “closed boundary around the system” that
Forrester put forward explicitly in this article to describe his approach. There are no
causal links coming from outside. There is a hint of “roundness” to the picture that would
be characteristic of such a closed causal boundary and the feedback structure it forces.
• Second, the dynamics generated by the model come from interactions of the variables
inside the model boundary. There is no declining market cap coming from outside to
inhibit growth. We see that the self‐contained loop structure shown in Figure 3(a) is
sufficient, by itself, to generate the observed dynamics.
• Third, we see that Forrester designed the model to tell the endogenous story of
declining sales in the most vivid way possible: he put the company in a potentially
infinite market. One might well argue that no company exists in such a market, and so
the model is unrealistic and invalid. But the potentially infinite market takes away all
possibility that the declining sales can trace to anything other than an endogenous
source. This is modeling for endogenous insight and understanding.

Think how extraordinarily difficult, if not impossible, it would be to reach these

potentially crucial policy‐related conclusions without an endogenous perspective or, to
put it more forcefully, without an endogenous bias in one’s point of view.

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 225

This endogenous perspective is neither well understood nor accepted by many

practitioners and scholars. The second classical Forrester example provides a dramatic
illustration.

Urban Dynamics
The 1960s was a period of urban renewal in the U.S.A. Many policies were implemented
in efforts to halt central city decay and return old cities to the economic vibrancy of their
earlier days. Forrester engaged in conversations with a group of urban experts including
John Collins, former mayor of Boston and past president of the National League of Cities.
The book Urban Dynamics, and the model on which it was based, emerged from those
conversations (Forrester, 1969, pp. vii–x).
Figure 4, reproduced directly from the book, shows how Forrester chose to frame the
study. The perspective he took, and the model he built, viewed an archetypical city in a
large and dynamically uninteresting environment. Significantly, the model did not
include suburbs around the city, or the transportation networks linking it to its
environment (and it was criticized for such glaring omissions). In fact, the environment
around the city generated no dynamic influences whatsoever on the modeled city, and the
city was presumed to be small enough in the world setting not to affect its environment. In
reality, there might be external economic cycles, political and social movements, cultural
changes, and so on, affecting the city and its environment, but Forrester chose explicitly to
ignore all those exogenous dynamics. The dynamics of interest were urban dynamics
relative to the environment outside the city. They were to be insistently, undeniably
endogenous dynamics generated by the city itself.
To give a feel for what the Urban Dynamics model looks like, Figure 5 shows the
structure of URBAN1, a simplified three‐stock version created by Alfeld and Graham
(1976).4 To build toward understanding Forrester’s more complex structure, Alfeld and
Graham formulated URBAN1 to have single stocks of business and housing structures,
where Forrester had three‐stock aging chains of new structures, mature structures and

Fig. 4. Forrester’s picture of the conceptual setting of Urban Dynamics (Forrester, 1969, p. 15)

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
226 System Dynamics Review

Fig. 5. The stock and flow feedback structure of URBAN1 (Alfeld and Graham, 1976). Note no exogenous
dynamic influences: the behavior of the model is generated completely by the endogenous interactions
represented in the diagram

aging structures. Forrester also disaggregated population into three linked stocks of
underemployed, skilled workers, and managerial professionals, while URBAN1 contains
just a single stock of people. Forrester’s model had about 150 equations; URBAN1 has 21.
In spite of the simplifications, the character of URBAN1 matches the Urban Dynamics
model and it provides a marvelous introduction to the insights Forrester obtained.
Figure 6 shows some of the key variables from the base run of the model.

Fig. 6. Dynamic behavior of URBAN1, showing endogenous growth, stagnation, and decay of population and
businesses, and the dramatic rise of unemployment in the built‐up city

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 227

The figure shows a growth phase with low unemployment (indicated by the labor‐to‐job
ratio), which turns rather quickly into a no‐growth stagnation phase, followed by a long
period of decline in population and business structures. During the stagnation phase,
unemployment rises and remains high for the rest of the simulation.
Like the larger Urban Dynamics model, URBAN1 captures endogenously the dynamics of
mature cities.5 The emergence of urban stagnation, the rise in unemployment, and the onset
of urban decay come about without any external influences. Furthermore, the behavior of
these urban models is astonishingly resistant to parameter changes. Figure 7, for example,
shows what happens when each business in the city is magically able to hire 30 percent
more people at time 40, near the beginning of the urban decay phase, an incredibly
successful jobs program. There is a sudden drop in unemployment, followed by a rise in
population and a consequent rise in unemployment to the same level reached in Figure 6.
The jobs program works in the short run and is then completely compensated for by a short‐
term rise in the attractiveness of the city. One can also note that the city has fewer business
structures by the end of this jobs program run: the period of lower unemployment raises
wages and makes the city relatively less attractive for new businesses.
Once again, we see a number of crucial insights stemming from an endogenous
modeling approach and an endogenous perspective on socio‐economic dynamics:

• The dynamics of the city in these urban models do not depend upon dynamics outside
the system boundary, in what Forrester conceptualized as a “limitless environment”.
All dynamics are generated from within. The suburbs are not causing urban decay; the
cities in these models are doing it to themselves.
• The possibility that urban growth, stagnation, and decay have sources that trace to
perceptions, interactions, and forces inside the city itself would never have emerged
had Forrester expanded the system boundary to include suburbs, national urban policy,
or national economic dynamics. The endogenous point of view was crucial for the
insights we can derive from URBAN1 and the Urban Dynamics study. In fact, the
choice of system boundary narrower than most would have taken proved to be the key
to deep urban dynamics insights.

Fig. 7. Behavior of URBAN1 when a jobs program is instituted at time 40

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
228 System Dynamics Review

• Forcing dynamics to be endogenous forces causality in these models to be circular.

Feedback loops in these models are consequences of the assumption of a closed causal
boundary. Thus the endogenous point of view provides the essential perspective capable
of illuminating crucial compensating feedback effects that can conspire to defeat favorite
policies.
• Feedback loops become crucially important for understanding urban policy. For
example, the feedback loops containing the labor‐to‐job ratio here (see the upper right
of Figure 5) completely compensate for the simulated jobs program and render it
useless in the long run.

How might thinking exogenously have affected conclusions emerging from these urban
models? Consider just one: suppose, as a number of critics of various system dynamics
studies have suggested, we chose to use time series data for urban population projections.
Suppose the data‐based projections were very carefully developed by sophisticated statistical
tools and econometric methods, and suppose those projections were fed into URBAN1 in
place of the endogenous stock of population. To make the implications easiest to see, let us
suppose the base run of the model looks just as it did in Figure 6. What would Figure 7 look
like? Sadly, the population would not show a bump as it did in Figure 7 because the
sophisticated exogenous time series data would not be influenced in the slightest by the
changing conditions in the model. We would not see the compensating urban migration
effects we see in Figure 7, and we would miss the crucial conclusion that population and
business construction dynamics would naturally compensate for the jobs program. We’d
probably think it was a long‐term policy success, and we would be dramatically wrong.
Perhaps that conclusion is obvious here. We see the compensating loops in the upper
right of Figure 5 rather easily, and we can easily understand the resulting dynamics and
gain insight. But it may come as a surprise to learn that experienced modelers and some
critics have been seduced by sophisticated statistical methods to think that time series
data would improve the validity of dynamic models of social systems. The follow‐on work
to World Dynamics (Forrester, 1971) and Limits to Growth (Meadows et al., 1972) provide
an instructive example.

A brief observation from World Dynamics and Limits to Growth

In the decade following the publication of these model‐based studies, many scholarly
centers around the world worked to contribute model‐based insights to the “global
problematique” made famous by the Club of Rome. In a retrospective on those efforts,
Forrester (1982) was forced to note rather critically that some of them held population
exogenous, fed into the models as time series data, using the best estimates available for
population data from sectors all around the world. He wrote to correct that tendency:

Many recent world models, by letting population be exogenous, lose feedback from
other variables back to population, and thus leave out the central dynamic factor
driving world growth. The essence of world modeling is to find the high‐leverage
policies that will stabilize population in time to retain a satisfactory quality of life. …
If population is put back as an internal dynamic variable, more study will then be
addressed to demographic change and to the underlying force that is threatening to
drive the world into physical and social limits. (Forrester, 1982, p. 102).

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 229

For example, improving health conditions on a global scale would lower infant mortality
and increase average lifespans, thus increasing for at least two or three generations the
population growth rate, resulting in a larger population to sustain materially and feed.
Continuing, he expressed his thoughts on the strengths of the system dynamics
approach, mentioning specifically endogenous dynamics:

Perhaps it is time to reintroduce system dynamics into world modeling: it lends itself
to communicating with the public, dealing with long time horizons, choosing the
appropriate level of aggregation, emphasizing policy choices, making all the variables
endogenous,6 joining the arena of political controversy, and drawing on the rich and
diversified mental database. (Forrester, p. 1982, p. 104).

In Forrester’s view in this global modeling context, endogeneity and the feedback
perspective it forces are crucial to responsible discussion of the global problematique and
the reliability of potential national, regional, and global policy directions.

Endogeneity and agency

We turn now to the question of agency in system dynamics: Who are the actors in the
dynamics of a complex system and how do their perceptions, pressures and policies
interact? Are you and I, or the groups we represent, part of the endogenous system
structure responsible for the system behavior we perceive? Are we parts of the problem, or
parts of the solution, or merely bystanders watching difficult dynamics play out over time?
The quotations at the beginning of this article illustrate that the question of the agents of
endogenous dynamics has been with us for centuries. It is present in a number of current
policy areas and is likely to bedevil future generations as well. Wisdom appears to side
with inclusivity: certainly Cassius, Disraeli, and Pogo would have us believe that we are
endogenous actors in the grand issues we face.
But others might have us believe differently. The following examples probe exogenous
and endogenous points of view on problems of our time and the question of exogenous or
endogenous agency in the problem dynamics. We’ll conclude with an important framework
for viewing such complex issues and a rich understanding of what systems thinking and
system dynamics can contribute. For illustration, we will begin on a very homey note about
personal relations, and move to big issues: global climate, terrorism, and floods.

Personal relations
In the Borton Primary Magnet School in Tucson, Arizona, systems thinking has been
enthusiastically integrated into classwork and organizational practices. I visited there last
year and met three first‐grade boys and their gifted teacher, Kathy Lohse. They enchanted
me with a personal systems story.7 The boys, about 6 years old, are best friends, but kept
finding that they got into heated arguments on the playground, and somebody always got
his feelings hurt. Until they took a systems view, the boys saw only random disconnected
events. Ms Lohse helped the boys see the unhappy reinforcing loop shown in Figure 8. It
is a very simple view of undoubtedly complicated three‐way interpersonal dynamics, but
the boys drew it themselves (or so it seemed under the guidance of their gifted teacher)

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
230 System Dynamics Review

Fig. 8. A reinforcing feedback loop drawn by three first‐grade boys that helped them take an endogenous view
and see their own roles in their repeated playground arguments

and they owned it. Realizing that mean words hurt feelings and set up the likelihood of
more mean words enabled the boys to think about ways they could break the spiral.
The boys have internalized at least some of the lessons here. At one point two of them
got into an argument cleaning up the classroom, tussling over putting away the same
chair. The third interrupted them: “Guys! Remember the loop!” And the two fighting over
the chair backed off, each offering it to the other to take care of: “You take it.” “No, you … .”
They are beginning to see themselves in the dynamics they experience. To paraphrase
Disraeli, sometimes we might tend to think we are creatures of circumstance (an exogenous
view), but wisdom frequently suggests circumstances that surround each of us are creatures
we help to create (an endogenous, feedback view).

Global climate
Various groups in the U.S.A., Europe, and the rest of the world are engaged in
occasionally heated conversations about global climate change. The data about global
average temperature seem not to be at issue. Figure 9 shows 100 years of global
temperature (relative to the 1961–1990 average) measured in degrees Celsius per year. The
figure shows that the global average temperature has been rising steadily throughout the
20th century, and rising particularly sharply since 1970.

Fig. 9. One hundred years of global average temperature shown as deviations from the average temperature
from 1961–1990, measured in degrees Celsius8

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 231

That sharp rise coincides with a sharp rise in the global average atmospheric
concentration of carbon dioxide, a significant heat‐trapping gas (see Figure 10).
That is where the controversy begins. Some argue that human activity is at least partly
responsible for the rise in global temperature because it is human production and use of
fossil fuels that have pushed CO2 levels to dramatic heights. Others argue that current
changes in global CO2 concentrations and temperature are simply part of a natural cyclic
phenomenon that has been going on for millennia. Figure 11, for example, shows 400
thousand years of data on temperature and CO2 and methane concentrations inferred from
Arctic ice core data. In each time series, the data show a vivid cycle with a period of about
100,000 years. Some argue that we are simply in the most recent of these natural cycles.
Thus there is an exogenous point of view on global warming, maintaining that current
conditions are simply part of an exogenously generated cycle, and an endogenous view,
holding that human activity in the form of the burning of fossil fuels is exacerbating
temperature increases.
The exogenous view has some support in the mechanisms known as Milankovitch
cycles (source: http://en.wikipedia.org/wiki/Milankovitch_cycles). Figure 12 shows the
data supporting the theory that structural aspects affecting the intensity of solar radiation
reaching earth are responsible for long‐term ebbs and flows in the earth’s incoming solar
energy.
As Figure 12 illustrates, these Milankovitch cycles can apparently produce very long‐
term oscillatory patterns in global temperature, shown in the figure as stages of glaciation.
It is worth noting that both the exogenous and endogenous views on global climate
change have a feedback structure underlying at least some of the dynamics (see Figure 13).
The water and carbon cycles are influenced by atmospheric and ocean temperatures, and
they in turn feed back to influence global temperatures via reflective effects from clouds
and ice and heat‐trapping effects from water vapor and CO2 concentrations. Figure 13
shows four external influences on interactions in the global climate system. Those who
argue for a human role in rising global temperatures would emphasize fossil fuel use and
human production of aerosols and greenhouse gases. Those who argue for a long‐term
natural cycle similar to the Milankovitch theory would de‐emphasize those external forces
and instead emphasize variations in solar energy reaching earth.
Thus, while balancing and reinforcing feedback interactions are present in all serious
views of global climate dynamics, the global warming debate comes down to questions of
agency and the implications for human action. Table 1 summarizes the points of view.

Fig. 10. Global average concentration of atmospheric carbon dioxide over the past 1000 years, showing a
dramatically sharp rise since 1900

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
232 System Dynamics Review

Fig. 11. Global average temperature and atmospheric CO2 and methane concentrations over the past 400,000
years, as inferred from Arctic ice core data

Fig. 12. A million years of hypothesized data used by Milankovitch to suggest that the Earth’s climate variations
can be traced to structural aspects of earth’s orbit and solar intensity (source: http://www.eoearth.org/article/
Milankovitch_cycles)9

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 233

Fig. 13. An overview of the water and carbon cycles and their feedback effects on global temperature, showing
four potentially exogenous influences (sources: Bernstein et al., 1994, and related work of the author; see also
Fiddaman, 2002)

Table 1. Summary of exogenous and endogenous agency perspectives on global climate change

Global climate Perspective Policy implication

Exogenous view We are in the warm phase of a Adapt to the inevitable.

100,000‐year cycle caused by
exogenous, structural characteristics
Endogenous view Human activity is exacerbating Alter human habits to minimize
the natural cycle. the coming tragedies.

Floods
The destruction caused by devastating floods along rivers and coastal waters provides
another vivid contrast of exogenous and endogenous agency. Figure 14 shows the increasing
costs of flood damages in the U.S.A. over the course of the 20th century.

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
234 System Dynamics Review

Fig. 14. Annual costs of flood damages in the U.S.A. over the 100 years from 1903 to 2003, in constant (1913)
dollars, showing an apparent tendency for costs to be increasing over time (sources: Deegan, 2007; NOAA)

Fig. 15. An exogenous perspective on floods and flood damage, suggesting the reasonable conclusion that the
main causes of flood damage are the random severity and frequency of floods

One obvious perspective on the increasing costs would hold that over the 20th century
floods in the U.S.A. were becoming more frequent, or more severe, or both. Figure 15
captures that essentially exogenous perspective on flood damage.
However, Deegan’s (2007) extensive analysis suggests another explanation: an endogen-
ous view of the dynamics of flood damage that takes account of the human role in creating
property vulnerable to flood damage. Citing Blaikie et al. (1994), he points out that “disaster
occurs when hazard meets vulnerability”. He then traces the dynamics of vulnerability to the
interacting actions of the capacity of the local environment to withstand floods, development
pressure, property tax needs, perceived risks of development, moral hazard, policy
entrepreneurs, and other people pressures. Figure 16 shows an overview of the feedback
structures he hypothesizes to be at work in the dynamics of flood‐related property damage.

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 235

Fig. 16. Endogenous view of the dynamics of flood‐related damage (source: Deegan, 2007)

Fig. 17. Dynamics of five indicator variables (Deegan, 2007; ch. 5) showing what happens when the model is
disturbed by five identical floods spaced 10 years apart. The damage (bold curve on bottom) resulting from the
five floods is different each time, dramatically illustrating the dominant role of human agency in creating
vulnerable property, reducing the local environmental capacity to withstand floods, and increasing the resulting
property damage from floods

Deegan’s simulation experiments are telling. In the base run, he simulates five identical
floods 10 years apart. The behaviors of several key indicators are shown in Figure 17. The
important observation is that the property damage (thickest line) is not the same after every
flood. The causes for variable damage over time trace to the endogenous feedback structures
outlined in Figure 16 which generate the dynamics of vulnerable property (top curve).

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
236 System Dynamics Review

Table 2. Summary of exogenous and endogenous agency perspectives on flood damage

Flood damage Perspective Policy implication

Exogenous view Floods happen sometimes; the greater the When floods happen to occur,
flood, the worse the damage recover and rebuild
Endogenous view Damage occurs when hazard meets Recognize human role in damage.
vulnerability; vulnerability is a result of Work with stakeholders to minimize
people policies vulnerabilities

Fig. 18. Time series data on terrorist incidents worldwide. Colors distinguish regions in which attacks
occurred. The dominant region in the 1980s was South America; the later post‐2000 emerging dominant
regions are the Middle East and South Asia (source: http://www.start.umd.edu/gtd/features/GTD-Data-
Rivers.aspx)

Again, we have open‐loop and feedback perspectives on the dynamics of flood damage.
Table 2 summarizes these exogenous and endogenous points of view.
We should observe that only in the endogenous point of view is there the empowering
perspective that human behavior can have a significant policy role that could minimize
future flood‐related damage.
Terrorism
The dynamics of terrorism dominate the concerns of today’s world. Figure 18 shows data
on incidents in various regions of the world from 1970 to 2005 compiled by Gary Lafree
and colleagues at the START Center at the University of Maryland.
Even a very brief search on the Internet can reveal a host of theories about causes of
such terrorism incidents. Here is an arbitrary random sample:

• Ethnicity, nationalism/separatism, poverty and economic disadvantage, globalization,

(non)democracy, Western society, disaffected intelligentsia, dehumanization, and
religion (excerpted from http://www.meteck.org/causesTerrorism.html).
• Marginalization, ethnicity and nationalism, religion, cultism, free flow of weapons,
training of non‐military personnel, no pure democracy present (excerpted from http://
www.helium.com/).
• Belief causes terrorism (http://nobeliefs.com/terrorism.htm).

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 237

• Foreign domination and control of Muslim resources, hatred of the Western way of life,
alienation, poverty and illiteracy, moral decadence of the West, the West’s support for
Israel. (excerpted from http://www.wnd.com/news/article.asp?ARTICLE_ID=53842).
• Economic deprivations, political injustices, foreign occupation and denial of funda-
mental rights, including the right to self‐determination (“World must cooperate against
terrorism: Qureshi”, 10 June 2010, archive of http://www.dailytimes.com.pk/).
These examples and untold numbers of media accounts suggest that people have a
tendency to think about terrorism in much the same open‐loop terms that dominate most
of our daily intuitive thinking. For the most part, the language focuses on exogenous
sources and pressures. To be sure, one can begin to see interacting influences in words
like “marginalization” and “foreign domination”, but explicit loops of causal influence are
extremely rare in online and printed analyses.
Consider how different a feedback view of the dynamics of terrorist cells might look. In
Figure 19, the terrorist group grows as new recruits are attracted to it because of the zeal they
feel for their cause. The group experiences losses as some of its members are killed or defect
because of the dangers of terrorist actions. Those actions (suicide bombings, raids,
improvised explosive devices, and so on) lead to efforts to suppress terrorist activities and
groups, which can lead to losses from terrorist groups but can also inflame passions, make it
easier to recruit new terrorists, and contribute to peripheral support. That support can
increase funding and help to generate even more terrorist actions.
Of particular interest in an endogenous feedback view like Figure 19 is the stock called
“martyrs to the cause”. It is increased by terrorist deaths and decreases only when those
deaths are no longer remembered. It has a very long time constant. Recall that the split
between Shia and Sunni Muslims dates from 632, the year the prophet Muhammad died and
a disagreement developed over who should succeed him. That is more than 1300 years ago,

Fig. 19. A feedback view of the dynamics of a hypothetical terrorist cell somewhere in the world, or perhaps an
aggregate of linked terrorist cells10

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
238 System Dynamics Review

Table 3. Summary of exogenous and endogenous views of terrorism

Terrorism Perspective Policy implication

Exogenous view Violent forces exist that threaten others, Defend to prevent harm; attack to weaken
and they are growing or eliminate the violent forces
Endogenous view Violent forces interact with defenses Defend to prevent harm; minimize
and attacks to create the rising behaviors that create nasty reinforcing
tensions we observe loops; maximize creation of beneficial
reinforcing loops; work toward cross‐cultural
understanding

and it continues today in modern‐day Islam. The Armenian Genocide (a term disputed by
modern‐day Turkey) occurred between 1915 and 1917, more than 90 years ago, yet is a source
of current impassioned debate. William Wallace, a leader in the Wars of Scottish
Independence, was found guilty of treason by the English and was executed in 1305. About
700 years later, in 2002, he was ranked #48 as one of the 100 greatest Britons in a poll
conducted by the BBC (http://en.wikipedia.org/wiki/William_Wallace). These few examples
suggest that reinforcing loops containing “martyrs” are extremely worrisome sources of very
long‐term dynamics.
Thus we have mixed exogenous and endogenous (feedback) views of terrorism. Table 3
sketches their characteristics and implications.

The great insight of servomechanisms engineering

As we saw in the first part of this paper, the early writings of Forrester suggest that the
endogenous point of view is a crucial aspect of the system dynamics approach. Also, we
have identified in the various examples described above at least some of the important
societal implications of taking an exogenous versus endogenous perspective.
However, the importance of endogeneity can be very local and visceral, as this story
told by Forrester from his student days suggests:
One time we were making feedback control systems with some high‐powered
applications—I think it was a 10‐horsepower motor with a hydraulic control system to
drive probably some kind of military gun mount. I remember one night I was working
with it, and something went wrong. It had become unstable, and it began to go back and
forth at the maximum speed that the 10‐horsepower motor would drive it. Some of the
hydraulic lines had broken, and it was spraying oil into the air, and I was trying to get it
stopped. As I rushed over to try to turn it off, I slipped in the oil on the floor. What I
remember is seeing the rainbows in the oil spray up against the lights … which is a lesson
on oscillatory behavior.11
A lesson indeed. The story reminds us that Forrester was a servomechanisms engineer
by training and inclination. He studied electrical engineering and servomechanisms
control with Gordon Brown, who founded MIT’s Servomechanisms Laboratory in 1940
(Richardson, 1991). Brown surely had experiences like Forrester’s oily slip in mind when
he wrote in his early servomechanisms text:

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 239

A closed‐loop control system is thus an error‐sensitive system and, being such, it

acquires certain peculiarities and idiosyncrasies which, in large measure, are the
reasons for this book. (Brown and Campbell, 1948).
The story and the excerpt from the introduction to the servomechanisms text also remind
us that “closed loop control systems”—feedback systems—often behave in counterintuitive,
unpredictable ways, as the young servomechanisms student Forrester was finding out. At the
heart of that unpredictable behavior lies the great insight of servomechanisms theory, and by
extension, system dynamics: the act of trying to govern/manage/control generates system
dynamics of its own.
The military gun mount goes haywire when its hydraulic control system overcorrects; a
muscle trembles trying to hold a heavy weight; the speed of a steam engine under the
control of a centrifugal governor oscillates as it “hunts” for the set‐point; the inflation rate
in the U.S. economy similarly oscillates under the efforts of the Federal Reserve Open
Market Committee to keep it and the economy on an even and productive keel; the efforts
of a government to dictate to its citizens generates rebellion; and so on. In each case, the
act of control generates dynamics of its own.
That insight—and it is indeed a deep, profound insight—underscores the importance of
the endogenous point of view in understanding the dynamics of social systems.

Choosing points of view

As we think about the way natural and social systems play out over time, we have a choice
of perspectives. We can view dynamics as arising largely from exogenous forces outside our
purview and control, or we can see ourselves and the decisions we make as part and parcel
of the ebb and flow of things. It also seems reasonable that reality has a similar “choice”: the
dynamics we are thinking about might come in reality from largely exogenous forces, or
from predominantly endogenous interactions. Thus we have a kind of two‐by‐two table, the
X/N matrix in Figure 20, a distant cousin of the Taylor–Russell diagram.
Figure 20 suggests four possibilities. In two of the cells we are “right”. The cell at the lower
left of Figure 20 represents those situations in which decision makers see the phenomena as

Fig. 20. The X/N matrix: eXogenous and eNdogenous perspectives contrasted with their corresponding “true”
states of affairs that are mainly exogenous or endogenous, showing four possible (idealized) perception/reality
combinations and their implications

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
240 System Dynamics Review

largely exogenously caused, and in reality that is essentially the true situation. In the upper
right, decision makers are taking an endogenous point of view of what are, in reality,
essentially endogenously generated dynamics. The other two cells represent situations in
which our perspective does not match reality very well, analogous to Type I and Type II
errors in statistics. Perspectives and reality probably do not fall neatly into such discrete
boxes, so we should think of these as something like “ideal types”.
In the lower left cell (exogenous view of exogenous phenomena), we see correctly that
there is little that can be done to alter the future state of affairs, so our best course is to
accept our fate, predict, and prepare for whatever we believe is coming. The expressions
on the faces in the table suggest that even though this is a correct perception, we probably
are not very happy about it.
In the upper left cell (endogenous view of exogenous phenomena) we are trying to find
endogenous understandings but we will fail in such attempts because the situation is
predominantly determined externally. So we are wrong, and our efforts at endogenous
explanation and understanding are misguided. Yet for this author and perhaps most of us,
the effort to find some endogenous aspects here give us hope for understanding and
control (the “illusion of control”) and probably makes us feel better than we do in the
lower left. So the faces are still mixed, but happier.
In the upper right (endogenous view of endogenous phenomena), we are taking the
correct perspective for the situation, and we are potentially empowered by it. The
situation is best understood in feedback terms, with the possibility of human agency in
contributing to problems and solutions. We are seeing the endogenous aspects of
corporate conditions, urban dynamics, global climate, or terrorism and the like from an
endogenous perspective, and we have the chance of influencing wisely how things will
play out over time. That cell merits three very happy faces.
In contrast, the cell at the lower right (exogenous perspective on endogenous
phenomena) is a dismal prospect, characterized by three very unhappy faces. Here we
take the point of view that circumstances are largely caused by external forces, when in
fact there are essentially correct endogenous, empowering explanations. In this cell we are
destined to be confused and misguided, and we will misguide others.
If one were to recast Figure 20 as a decision tree, the decision to take an endogenous
point of view in all circumstances would have the highest net payoff, at least in terms of
happy faces and the real feelings they represent. An endogenous point of view is
potentially empowering, and that feels good to us.

The foundation of systems thinking and system dynamics

We began this exploration with the early thoughts of Forrester on the essential nature of
the field he created. Through those early writings and a number of wide‐ranging examples
we have plumbed the nature of exogenous and endogenous points of view. We are now
ready to try to expose the implications of these ideas and examples. The implications now
appear to enable both a tight characterization or even definition of the field of system
dynamics and perhaps an accurate characterization of systems thinking in general.
Our first implication is, by now, rather well established:

The endogenous point of view is a crucial foundation of the field of system dynamics.

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 241

While we know that there are legions of systems thinkers, and they have their own
special interpretations and contributions,12 it now seems very likely that they draw their
deepest fundamentals from what we have been investigating. Could we capture a key
aspect of all of them in the following sweeping characterization?

Systems thinking is the mental effort to uncover endogenous sources of system behavior.

An appealingly concise characterization of the field of system dynamics results:

System dynamics is the use of informal maps and formal models with computer
simulation to uncover and understand endogenous sources of system behavior.

A potential corollary is important to the understanding of our field: analyses that do not
seek to uncover endogenous sources of system behavior are not really system dynamics
applications. There may be formal models in a study, there may be stocks and flows, there
may even be feedback loops, but without a strong endogenous character the analysis is
missing a defining aspect of the system dynamics approach. Such analyses will be prone
to miss crucial ramifying effects and intricate compensating mechanisms that an
insightful endogenous perspective could reveal.
From this point of view, we can say quite clearly what system dynamicists do:

System dynamics practitioners use systems thinking, management insights, and

computer simulation to
• hypothesize, test, and refine endogenous explanations of system change, and
• use those explanations to guide policy and decision making.
This characterization of the system dynamics approach is true to the writings of the
founder of the field and is reflected in the best work of its practitioners.13

Notes

1. “Endogenous” (from the Greek fragments “endo” and “gen”, meaning “inside” and
“production”) refers to an action or object coming from within a system. It is the
opposite of “exogenous”—something generated from outside the system (http://en.
wikipedia.org/wiki/Exogeny). Systems thinking and system dynamics center on
endogenous phenomena, often encapsulated as “system as cause”.
2. Forrester’s use of the term “closed” means “causally” closed. His use is different from
the notion of a closed system in general systems theory, which refers to a system that
is “materially closed”, that is, does not exchange material or information with
anything outside the system boundary. Forrester’s “closed boundary” systems are, in
general systems theory terms, “open systems” because they include little clouds
representing sources and sinks of material outside the system boundary (see
Richardson, 1991, p. 298).
3. In practice, confronted with a view like the left side of Figure 2 one strives to expand
the system boundary to draw in the influences that initially seemed exogenous and
create a rich endogenous view.
4. Alfeld and Graham’s Introduction to Urban Dynamics remains one of the great
teaching texts in the field. It builds from simple building blocks up to the complexity

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
242 System Dynamics Review

of Forrester’s model, carrying the reader’s understanding along in the process.

URBAN1 appears about halfway through the text.
5. For typical long‐term population data on three U.S. cities, see Ghaffarzadegan et al. (2011).
6. To my knowledge, this is the first time Forrester used the word in print.
7. I am indebted to Tracy Benson of the Waters Foundation for taking me to Borton,
introducing me to teachers and children there, and making it possible for me to talk with
Kathy Lohse and these three boys and hear their story. See a video of the boys explaining
their loop at http://www.watersfoundation.org/webed/mod5/examples/mod5‐5‐11.html.
8. Source of all graphs in Figure 9–11: http://www.landcareresearch.co.nz/research/
globalchange/climate_change.asp#TempChange. For more graphs, data sources, and
analysis see http://en.wikipedia.org/wiki/Temperature_record_of_the_past_1000_
years>.
9. Figure 12 definitions: Precession—a gradual shift in the direction of the Earth’s axis,
which exhibits overlapping cycles from 19,000 to 24,000‐year periodicities. Obliquity—
the plane of the Earth’s orbit around the sun, which shifts in cycles of about 40,000‐year
periodicity. Eccentricity—the “ovalness” of the earth’s orbit, which shifts from more
circular to more elliptical and back in overlapping cycles from 95,000 to 400,000‐year
periodicities. Solar forcing—heat energy from the Sun reaching Earth, which exhibits
long‐term oscillatory patterns.
10. Two of the causal links in this diagram are negative links, and are marked; all the rest
are positive links. Some of the loop polarities are given to illustrate that this diagram
has an overabundance of reinforcing (R) loops.
11. An excerpt from Jay’s Stories, an unfinished, unpublished collection of stories Forrester
has told and retold over the years, collected in interviews by the author in 2006–07.
12. See, for example, Richardson et al. (1994).
13. See, for example, http://www.systemdynamics.org/AwardRecipients.htm for a list of
winners of the Forrester Award.

Biography
George P. Richardson is O’Leary Professor of Public Administration Policy in the
Rockefeller College of Public Affairs and Policy at the University at Albany, and affiliated
Professor of Informatics in the College of Computing and Information. He is the author of
Introduction to System Dynamics Modeling with DYNAMO (1981), Feedback Thought in
Social Science and Systems Theory (1991, 1999), both of which were honored with the
System Dynamics Society’s Forrester Award, and the edited two‐volume collection
Modeling for Management: Simulation in Support of Systems Thinking (1996).

References

Alfeld LE, Graham AK. 1976. Introduction to Urban Dynamics. Pegasus Communications: Waltham, MA.
Bernstein DS, Richardson GP, Stewart TR. 1994. A pocket model of global warming for policy and
scientific debate. In Proceedings of the 1994 International System Dynamics Conference, Stirling,
UK, Vol: Organizational Environments, July 1994. System Dynamics Society: Albany, NY; 8–18.
Blaikie P, Cannon T, Davis I, Wisner B. 1994. At Risk: Natural Hazards, People’s Vulnerability, and
Disasters. Routledge: London.

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
 G. P. Richardson: Foundations of System Dynamics 243

Brown G, Campbell DP. 1948. Principles of Servomechanisms. Wiley: New York.

Deegan M. 2007. Exploring U.S. flood mitigation policies: a feedback view of system behavior. PhD
dissertation, Rockefeller College of Public Affairs and Policy, University at Albany–SUNY.
Fiddaman TS. 2002. Exploring policy options with a behavioral climate–economy model. System
Dynamics Review 18(2): 243–267.
Forrester JW. 1958. Industrial dynamics: a major breakthrough for decision makers. Harvard
Business Review 36(4): 37–66.
Forrester JW. 1961. Industrial Dynamics. Cambridge, MA: MIT Press. Reprinted by Pegasus
Communications, Waltham, MA.
Forrester JW. 1968a. Market growth as influenced by capital investment. Industrial Management
Review 9(2): 83–105 (now Sloan Management Review). Reprinted in Collected Papers of Jay W.
Forrester. Pegasus Communications: Waltham, MA.
Forrester JW. 1968b. Principles of Systems. Pegasus Communications: Waltham, MA.
Forrester JW. 1969. Urban Dynamics. Pegasus Communications: Waltham, MA.
Forrester JW. 1971. World Dynamics. Pegasus Communications: Waltham, MA.
Forrester JW. 1982. Global modeling revisited. Futures 14(2): 95–110.
Ghaffarzadegan N, Lyneis J, Richardson GP. 2011. How small system dynamics models can help the
policy process. System Dynamics Review 27(1): 22–44.
Meadows DH, Meadows DL, Randers J, Behrens III WW. 1972. The Limits to Growth. New York:
Universe Books.
NOAA. National Weather Service Hydrologic Information Center. Available: http://www.nws.noaa.
gov/oh/hic/flood_stats/Flood_loss_time_series.shtml [18 March 2011].
Richardson GP. 1991. Feedback Thought in Social Science and Systems Theory. University of
Pennsylvania Press: Philadelphia, PA. Reprinted by Pegasus Communications: Waltham, MA.
Richardson GP, Wolstenholme E, Morecroft JDW (eds). 1994. Systems Thinkers, Systems Thinking.
Special issue and conference edition of System Dynamics Review 1994; 10(2–3).

Copyright © 2011 System Dynamics Society Syst. Dyn. Rev. 27, 219–243 (2011)
DOI: 10.1002/sdr
Copyright of System Dynamics Review (Wiley) is the property of John Wiley & Sons, Inc. and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.