Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

GEO-SD303: Assignment 1

Saeed P. Langarudi∗

Autumn 2023

Requirements
• Please submit your assignment as an electronic, editable text file through
Mitt UiB.

• In your work, please include all the models or any other supplementary
materials developed for the assignment.

• In the homework assignments, you seldom (if at all) need to work with
causal loop diagrams (CLDs). Wherever you are asked to develop and
present a model diagram showing polarity of causal links or feedback
loops, that refers to the stock-flow diagram of your model. If a CLD
is needed, it will be mentioned explicitly.

• In this course, we use InsightMaker as our primary modeling software.


All the models must be developed in this software. Models in other
formats will not be accepted.


saeed.langarudi@uib.no

1
Introduction
During the lectures, we developed a model as an interpretation of Adam
Smith’s theory of converging rewards. In the current assignment, you are
expected to recreate the model, modify some of its assumptions and exam-
ine its behavior in response to the changes. The exercises help you learn
how a model can be calibrated to data. You will practice documenting and
reporting a model. You will also practice initializing a model in a steady
state (dynamic equilibrium) both numerically and analytically. Finally, you
will learn about the concept of first-order control and how to perform ex-
treme condition tests on a model. To complete the assignment, please do
the following exercises and answer the questions as concisely as possible.
Each exercise has 10 points.

Exercise 1
1. Using the equations provided in the lectures, replicate Adam Smith’s
converging rewards model in the software. Provide the link to your
model. Your model must

(a) have all the units correctly and consistent [2 points],


(b) include all the polarities (+/- signs for the causal links and feed-
back loops) in it diagram [2 points],
(c) include a very concise explanation for every equation (use the
"note" feature) [2 points].

2. Report your model equations in a way that is readable and accessi-


ble for a non-SD reader. You must use either integral or differential
equations. Each equation must have an explanation of its logic. This
explanation should be consistent with what you already inserted in
the notes of your model. Here, you need a little more elaboration.
However, try to limit each logical explanation to 50 words [4 points].

Exercise 2
1. Run the model with the original settings (dt = 0.25, method = Euler).
Visualize each model’s outputs and post the graphs for the key vari-
ables, i.e., workers and reward_per_worker. Make sure both models
behave exactly the same. If not, resolve the issues before you report

2
the results. Report any mismatch you discovered in the previous step
and explain its causes [5 points].
2. Run both models with a smaller dt, e.g., dt = 0.125, and compare the
results with the previous setting. Did you notice any difference? If yes,
what dt should be chosen? If not, what dt should be chosen? Report
your results using model output graphs and explain your observations
(less than 200 words) [5 points].

Exercise 3
1. Try to achieve a dynamic equilibrium 1 (steady state) by changing the
model’s constants (parameters). Report the parameter values (con-
stants and initial values) that keep the model in equilibrium [4 points].
2. How do the equilibrium parameter values change if the initial workers
is set at 90 (i.e., W(0) = 90) [3 points]?2
3. What did you learn about the system from this initialization exercise
(less than 200 words) [3 points]?

Exercise 4
1. Find an analytical solution that keeps the model in equilibrium regard-
less of parameter values. The analytical solution in this context means
that some constants (parameters including initial values) should now
be a mathematical (simple algebraic) function of other constants or
initial values. You need to write an equation for initial workers (W(0))
in terms of other model constants [5 points].
2. In the model, replace the value of W(0) with the analytical solution.
Run the model with this new formulation. Run the model again with
a different value of h. Compare the two runs and report it using a
comparative graph [3 points].
1
Dynamic equilibrium (steady state) is a condition in which all the variables remain
unchanged throughout the simulation period. This means that all the model variables
should appear as horizontal lines if you plot them over time (the variables are on the
y-axis while time is on the x-axis).
2
To answer this question, you need to repeat steps 1 and 2 above, but you have one less
degree of freedom this time. That is, you cannot play with W0 as part of the initialization
process. In other words, you have a constraint to satisfy while initializing the model in
equilibrium.

3
3. Explain your observation and what you have larned from this experi-
ment (less than 200 words) [2 points].

Exercise 5
1. Assume the model is in a dynamic equilibrium. Without using a com-
puter simulation, try to predict the behavior of workers in reaction
to a 20% step increase in natural_rewards (rewards of other occu-
pations). Draw the expected behavior of the key variables (workers
and reward_per_worker) from your mental simulation on a diagram.
Report and explain the diagram. Explain how each variable changes
in relation to changes to other model variables [3 points]3 .

2. In the model, change the formulation of natural_reward so that it


includes a step rise at time=2: natural_reward = 1 + STEP(0.2,
2). Report the simulation results for the key variables (R and W) using
comparative plots [4 points].

3. Does the behavior of the model comply with your expectations? In


other words, does your previous explanation of the model behavior re-
main valid? Explain any differences between the computer simulation
and your mental simulation in less than 200 words [3 points].

Exercise 6
1. The flow hiring in the model is formulated in a way that can accept
both positive values (workforce addition) and negative values (work-
force attrition). This is not a good modeling practice. Explain why [4
points].

2. Modify the model structure so that the positive and negative values of
hiring rate are separated. Save your model with a new name. Report
the changes you have made (also include a link to the new model). In
your report follow the standard model documentation practice. That
is, add documentation in the model itself as well as explaining the
equations in the assignment text using mathematical notation and
logical description [5 points].
3
This is an important system dynamics modeling skill you need to acquire. For that, I
recommend you carefully read chapter 16 of Jay Forrester’s seminal book: Forrester, Jay
W. 1961. Industrial Dynamics. Cambridge, MA: Productivity Press.

4
3. Confirm that the new model structure is equivalent to the original
model by expanding4 the hiring equation of the original model [3
points].

4. Confirm the similarity of the two models by repeating the simulation


runs you performed in Exercise 5. Report the simulation results using
model output graphs. If the results are not similar, try to find out
why and explain your findings [2 points].

Exercise 7
1. In the previous exercise, you created a modified version of the converg-
ing rewards model, assuming that the workforce outflow (attrition)
is constant. Explain why (not) this assumption is plausible [2 points]5 .

2. To understand the problem with the current assumption, do the fol-


lowing [2 points]:

• Run an extreme condition test by assuming:


(a) natural_rewards = 20
(b) reward_budget = 10
• Run the model and explain the dynamic behavior of the key vari-
ables focusing on what hiring and attrition flows are doing.

3. The problem you discovered here is due to the lack of a first-order


control on the attrition outflow6 . To resolve the problem, you should
assume the outflow is a direct function of the stock. The simplest form
is a linear function such as outflow[t] = a * stock[t], where 0 <
4
Expansion is a basic algebra technique to break down an equation. See this tutorial
for help, if needed: https://www.mathsisfun.com/algebra/expanding.html.
5
Please note that a plausible assumption is not necessarily a realistic one. Plausibility
of an assumption depends on its usefulness and robustness. An assumption is useful if it
serves the purpose of the model with a minimum level of complexity. An assumption is
robust if it can endure extreme condition tests. Therefore, to answer the question, you
need to see if the assumption is useful and robust.
6
As a general modeling rule, all outflows in a model should be controlled by their
corresponding stocks through a first-order control. First-order here means an impact
without any delay involved. That is, the stock should keep the outflow in check to ensure
that it goes to zero before the stock becomes negative. If the control is of higher order,
there will be delays in signals from the stock to the outflow. Therefore, the outflow might
keep going at positive rates while the stock is very close to complete depletion, risking it
going negative in the next time steps.

5
a < 1. The function could also be of nonlinear forms. Reformulate
the attrition rate so that it includes a first-order control. Save your
new model with a different name. In less than 200 words, report
your changes using the standard model documentation practices. In
your report, provide a logical explanation for your new formulation of
attrition [4 points].

4. Run the extreme condition tests of the previous exercise on the new
model and compare the two models’ results. Post model output graphs.
Were you able to resolve the extreme condition problem using the new
formulation [2 points]?

Exercise 8
Table 1 shows the data for average weekly earnings of construction workers
in the US between 2011 and 2021 in terms of 1982-1984 dollars (source: US
Bureau of Labor Statistics).

Table 1: Average weekly earnings of all employees, 1982-1984 dollars, con-


struction of buildings, not seasonally adjusted (USBLS)
Year Earning
2011 435.54
2012 440.22
2013 449.27
2014 453.97
2015 470.35
2016 476.47
2017 480.05
2018 491.73
2019 495.08
2020 502.38
2021 496.78

Do the following:

1. Insert data into the software. For that, you need to add a new con-
verter to the model. Replace its data points with the numbers from
Table 1. You should also adjust the simulation time to run from 2011
to 2021 [3 points].

6
2. Adjust the model parameters so that the variable R fits to the data as
closely as possible. Save the model with a new name and report the
link [2 points].

3. Report the results explaining the challenges you have faced and in-
sights you gained [5 points].

Exercise 9
1. Write a critical review of the model explaining how well it explains
the system’s dynamics. Comment on its validity. And, what factors
you consider in evaluating its validation. Limit your response to 500
words [6 points].

2. What feedback mechanisms could be added to the model to enhance its


validity and improve its explanatory power. Include a model diagram
supporting your explanation. limit the text to 500 words [4 points].

Exercise 10
Review the literature and find feedback theories that explain an industry’s
wage dynamics. Report at least one relevant system dynamics article. Ex-
plain how similar or different it is to Smith’s converging reward theory. Limit
your response to 500 words [10 points].

You might also like