MDA 5303: System Dynamics

Modelling

Lesson 2: Systems
Thinking, Feedback Structure,
and Policy Simulation

Lucy W. Mburu, Ph.D

 Systems Thinking
 A system perspective is focused on
patterns of behavior happening over
time, events and decisions
 A social context is a system with
components (influencing factors) and
sub- systems
 Seeing
“interrelationships”
among “systems” rather
than linear cause-and-
effect chains when
events occur
 Seeing “processes” of
change among “systems”
rather than discrete
2.2
 System Dynamics
 System dynamics is an approach to
understanding the behavior of complex
systems over time. It deals with internal
feedback loops and time delays that affect
the behavior of the entire system.
 What makes using system dynamics
different from other approaches to
studying complex systems is the use of
feedback loops and stocks and flows.
 The basis of the method is the recognition
that the structure of any system — the
many circular, interlocking, sometimes
time-delayed relationships among its
components — is often just as important in
2.3
 Feedback
 Feedback is a phenomenon whereby some
proportion of the output signal of a system
is passed (fed back) to the input. This is
often used to control the dynamic behavior
of the system.
 An example of a feedback system is an
automobile steered by a driver.

2.4
Four Key Patterns of Thought

 Dynamic thinking (graphs over time)

 Causal thinking (feedback loops)
 Stock-and-flow thinking
(accumulations)
 Thinking endogenously (system as
cause)

2.5
 Dynamic Thinking
 Dynamic thinking is concerned with
behaviors/ patterns “over time”
 Convert ambiguous, discrete events into
patterns of behavior
 Focus on “policy structure” rather than
individual decisions
 Define your problem in terms of graphs
over time
 Use “graphs over time” to perform
systems thinking and modeling
Current
weight

1 3 5 7 9 11 13 15 17 19
Time (week)

2.6
 Dynamic Thinking
 Dynamic thinking is concerned with
behaviors/ patterns “over time”
 Convert ambiguous, discrete events into
patterns of behavior
 Focus on “policy structure” rather than
individual decisions
 Define your problem in terms of graphs
over time
 Use “graphs over time” to perform
systems thinking and modeling
Current
weight

1 3 5 7 9 11 13 15 17 19
Time (week)

2.7
 Systems Thinking and
Dynamics
 How do we define problems in system
dynamics?
 Graph key variables
 Graph historical data (what happened)
 Graph predicted dynamics (what might
happen if nothing changes)
 Graph preferred dynamics (what we
prefer to happen)

2.8
Four Key Patterns of Thought

 Dynamic thinking (graphs over time)

 Causal thinking (feedback loops)
 Stock-and-flow thinking
(accumulations)
 Thinking endogenously (system as
cause)

2.9
Causal Loop Diagrams (CLDs)
 A CLD or causal map is a simple map
of a system with all its constituent
components and their interactions,
revealing the structure of a system.
 Depicts the structure of a system or a
set of causal relationships among
variables (activities) in a system
 Components of CLD are:
 Variables of a system
 Cause-and-effect relationships among
variables
 Feedback loop is a closed circuit of the
interconnection between variables
 Delay 2.10
 Polarities of Casual Links

The change of A leads to the

A + B change of B in the same
direction
The change of C leads to
C - D
the change of D in the
opposite direction

 Arrows indicate causal influence

 All words or phrases used to name
variables should be expressed as
quantities that have a clear sense of
increase or decrease
 No verbs should be used to name
variables – the actions are in the arrows
 Two types of causal-loop structures
 Reinforcing loops (a.k.a “R” or positive
loops) 2.11
 Feedback (Causal-loop)
Structure
 Reinforcing loops: “Destabilizing,
disequilibrating, growth producing, or
self-reinforcing”
+
Birth per year R Population
+ +

Birth ratio

 Balancing loops: “Self-governing, self-

regulating, self-equilibrating….. all implying
the presence of a goal…..”Death ratio

+
-
Populati B Deat
on h per
year
+

2.12
 Feedback (Causal-loop)
Structure
 Feedback Structure with both
Reinforcing and Balancing Loops
Death ratio

+ - +
R Population B Deaths per
Births per year year
+ +

+
Birth ratio

2.13
 Determining Loop Polarities
 Count the number of negative signs:
 An even number of “-” sign is a
reinforcing (R) loop
 An odd number of “-” sign is a balancing
(B) loop
 However, sometimes this role might not
be applied
 Fundamental Principle:
 Tell a self-reinforcing or
balancing/counteracting story for each
loop, and check that if the story
matches the loop polarity
2.14
 Drawing a CLD

1. Identify a systemic problem that is

chronic and recurring and has a
history and/or pattern.
2. Set a boundary and level of
simplicity
3. Identify key variables (activities)
that are significant events or
phenomena that jointly influence
the overall system.
 Use nouns or noun phrases
 Include “the level of,” “the amount of,”
“the number of,” or “the size of,” if
possible.
 Use a neutral or2.15positive term. Avoid
 Drawing a CLD
4. Begin with more interesting variable
(activity) and
 Work backward (begin with an event
and look for its cause)
 Work forward (begin with an event and
look for its consequence)
 Move back and forth (both directions).
5. Indicate “cause-and-effect”
relationship using a link and arrow.
 A departing point influences the
destination (A→B).
 No bi-directional arrow allowed (A↔B).
 Instead, use separate arrows (A→B,
B→A)
2.16
Drawing a CLD 5

E ffe c t 0
S

C ause 0

C ause 1

O
E ffe c t 1

2.17
 Drawing a CLD
6. Indicate the type of the relationship
between two variables on the link.
 + or S (same direction) means a
positive relationship between two
variables. If A increases, B will
increases; if A decreases, B will
decreases.
 - or O (opposite direction) indicates a
negative relationship. If A increases, B
will decrease; if A decreases, B will
increase.

2.18
 Drawing a CLD
7. Delay means that it takes long time
for an cause results in its effect.
 Depicted by || or “Delay” on the link
8. If a link between two variables is not
clear, redefine the variables or
insert an intermediate variable
between the two.
9. Keep revising iteratively; No CLD is
ever ending. Begin with a simple
CLD and try more challenging one
incrementally.
10. Label feedback loop with either “R”
for reinforcing or “B” for balancing
2.19
 An Example of CLD
 “Work backward” begins with the
important effect and then seeks
subsequent cause-effect chains.
 “Begin at the beginning” or work
forward
 “Go back and forth”

2.20
 An Example of CLD

B a n k F a ilu re
?

B a n k F a ilu re
O
?

B a n k F a ilu re
O
B a n k 's S o l v e n c y

B a n k 's S o l v e n c y

W ith d ra w a ls o f
Funds

2.21
 An Example of CLD
 Bank failure as a system problem
 First ask, “What caused the bank
failure?”
 Probably “bank’s solvency” is the
cause of “bankruptcy (bank failure).”
 What is the relationship between the
two variables? When bank’s solvency
is low, bank failure is more likely
(high). – or O

2.22
 An Example of CLD
 Then ask, “What caused bank’s
solvency?”
 Keep asking similar questions
B a n k F a ilu re
O

B a n k 's S o l v e n c y D e p o s ito rs '

C o n fid e n c e
O
B a n k F a ilu re
O

W ith d ra w a ls o f O
Funds O

D e p o s ito rs '
B a n k 's S o l v e n c y
C o n fid e n c e
O

W ith d ra w a ls o f O
2.23
Funds
 Feedback Loop
 A feedback loop is a closed circle of
interconnection between variables or a
series of mutual cause and effect (causal
relationships).
 Mutual causation is the simplest feedback
loop. A→B→A→B→…
 The loop is internally initiated by the
system and insensitive to environment.
 A closed circle requires that an activity
influences (comes back to) itself
eventually.
 Initial cause and ultimate consequence is
NOT distinguishable clearly.
 Given endless cause-effect
2.24 chains, such
 Feedback Loop
 “[E]verything is indeed connected to
everything else, so, in principle, it
doesn’t matter where you start”
(Sherwood, 2002: 128).
 “[C]ycling back means that what was
originally a cause is now suddenly an
effect” (Weick, 1979: 77).
 “[A]ny change made anywhere will
eventually itself be changed by the
consequences it triggers” (Weick, 1979:
77)

2.25
 Feedback Loop
 The interdependent relationships
in a structure (system) are more
important than initial cause or
starting point.
 Structure characterizes the
system and the starting point does
not matter.

2.26
 Reinforcing Feedback Loop
 A reinforcing or positive feedback
loop
 A feedback loop with even number
of negative signs (-) or O (i.e., 0, 2,
4, …).
 Depicted by “R” or “+” in the
center of a feedback loop
B a n k F a ilu re
 Self-reinforcing feedbackO
 Deviation amplifying feedback. O

No regulation or control
D e p o s ito rs '
 B a n k 's S o l v e n c y R (+ )
C o n fid e n c e
O

O
W ith d ra w a ls o f
2.27 Funds
 Reinforcing Feedback Loop
 “Reinforcing loops can be seen as
the engines of growth and
collapse” (Anderson & Johnson,
1997: 54).
 Destabilize the system and deviate
from its equilibrium
 Form a vicious circle or virtual
circle

2.28
 Balancing Feedback Loop
 A balancing or negative feedback
loop
 A feedback loop with odd number
of negative sign (-) or O (i.e., 1, 3,
5, …).  Goal seeking

B irth Depicted
S by “B” orfeedback,
“-” in the center
R  stabilizing
O
feedback,
P o p u la tio n
deviation-
S

B (-) counteracting, or
S

D e a th
self-refraining
2.29
feedback
 Multiple Feedback Loop
What if there are multiple feedback loops in
a system? Some loops are balancing, while
others are reinforcing
 If “loops are of unequal importance,” then
ask, “Which feedback loop is dominating
in the system?”
 Such dominating or the most important
loop will determine the fate of the system
in the end (Weick, 1979: 74).
 “[T]he greater the number of inputs to
and/or outputs from an element, the more
important that element is” (Weick, 1979:
75).
 “[C]lose loop that contains the greatest
2.30
 Multiple Feedback Loop
What if there is no dominant feedback loop? All
feedback loops are equally important.
 Count the number of balancing or negative
feedback loops
 Count the total number of negative
relationships (whenever it appears in a
loop)
 An odd number indicates a balancing or
deviation-counteracting system
 An even number indicates deviation-
amplifying system (Weick, 1979: 76).

2.31
 Multiple Feedback Loop

In a Loop In a System
Presence of a No influential loop
influential loop
# of negative Follow the # of negative or
relationships type of the balancing loops
between influential loop in a system
activities
If odd number, If odd number,
balancing loop balancing system

2.32
 How To Change a System?
 Add/delete activities (new system)
 Change in a relationship from –
(opposite) to + (same direction)
 Change in a relationship from –
(same) to - (opposite direction)
 Add/delete some relationships
 Shock (external input) will be
decayed without changing the
structure of a system

2.33
 References
 Anderson, Virginia, and Lauren
Johnson. 1997. Systems Thinking
Basics: From Concepts to Causal
Loops. Waltham, MA: Pegasus
Communications.
 Sherwood, Dennis. 2002. Seeing the
Forest for the Trees: A Manager.
London: Nicholas Brealey Pub.
 Weick, Karl E. 1979. The Social
Psychology of Organizing. 2nd ed.
Reading, MA: Addison-Wesley Pub.
Co. 2.34