D-4451-2 1

Modeling Exercises

Section 2

Population

Births Deaths

Prepared for the

MIT System Dynamics in Education Project

Under the Supervision of

Prof. Jay W. Forrester

by

Joseph G. Whelan

August, 1994

Vensim Examples added October 2001

Copyright © 2001 by the Massachusetts Institute of Technology

Permission granted to distribute for non-commercial educational purposes

 D-4451-2 3

Table of Contents

1. ABSTRACT 6

2. INTRODUCTION 7

3. EXERCISE 1: URBAN DYNAMICS 8

3.1 BACKGROUND OF THE URBAN SYSTEM 8

3.2 PRELIMINARY EXPLORATION OF THE URBAN SYSTEM 8

3.3 GENERAL DESCRIPTION OF THE URBAN SYSTEM 9

3.4 DETAILED INSTRUCTIONS FOR BUILDING THE URBAN GROWTH MODEL 10

3.4.1 THE “BUSINESS STRUCTURES” SECTOR 10

3.4.2 THE “POPULATION” SECTOR 11

3.5 EXPLORATION OF THE URBAN GROWTH MODEL 12

4. EXERCISE 2: MENTAL SIMULATION 14

4.1 EXPLORATION OF THE MENTAL SIMULATION MODEL 16

5. SOLUTION TO URBAN DYNAMICS EXERCISE 20

6. SOLUTION TO MENTAL SIMULATION EXERCISE 29

6.1 A NOTE ON TIME CONSTANTS 29

6.2 MENTAL SIMULATION EXERCISE SOLUTION 30

7. APPENDIX 36

7.1 EQUATIONS FOR URBAN GROWTH MODEL 36

7.2 EQUATIONS FOR MENTAL SIMULATION MODEL 39

8. VENSIM EXAMPLES 40

6 D-4451-2

1. ABSTRACT
This paper is the second in a continuing series of papers designed to enhance the
reader’s modeling skills. Modeling Exercises Section 2 contains two separate modeling
exercises. In the first exercise the reader will build a system dynamics model based on a
detailed description of an urban system. The reader will run the urban model under
varying conditions. The second exercise asks the reader to perform mental simulations on
a pre-built model, and to estimate the model behavior under different conditions. The
reader is expected to have prior experience with system dynamics, model building, and
mental simulation. Before beginning this paper, please review Modeling Exercises Section
1, which outlines the process of model building. 1

1
Joseph G. Whelan, 1994. Modeling Exercises Section 1 (D-4421), System Dynamics in Education

Project, System Dynamics Group, Sloan School of Management, Massachusetts Institute of Technology,

May, 1994, 37 pp.

Mental simulation is explained in the following papers:

Joseph G. Whelan, 1995. Beginner Modeling Exercises Section 2: Mental Simulation of Simple
Positive Feedback (D-4487), System Dynamics in Education Project, System Dynamics Group, Sloan
School of Management, Massachusetts Institute of Technology, July 27, 15 pp.
Helen Zhu, 1996. Beginner Modeling Exercises Section 3: Mental Simulation of Simple Negative
Feedback (D-4536), System Dynamics in Education Project, System Dynamics Group, Sloan School of
Management, Massachusetts Institute of Technology, January 28, 23 pp.
Alan Coronado, 1996. Beginner Modeling Exercises Section 4: Mental Simulation: Adding Constant
Flows (D-4546), System Dynamics in Education Project, System Dynamics Group, Sloan School of
Management, Massachusetts Institute of Technology, May 28, 27 pp.
Laughton Stanley and Helen Zhu, 1996. Beginner Modeling Exercises Section 5: Mental Simulation of
Combining Feedbacks in First Order Systems (D-4593), System Dynamics in Education Project, System
Dynamics Group, Sloan School of Management, Massachusetts Institute of Technology, May 28, 28 pp.
 D-4451-2 7

2. INTRODUCTION
Modeling is a crucial element of system dynamics. People rely on mental models
daily to make decisions without even questioning the model being used. Creating and
simulating models using STELLA or other modeling software2 allows the modeler to
solidify his mental models. The modeler thus observes the behavior of complex systems
where mental models are inadequate. Both mental and computer models can give wrong
answers. A computer will give an answer with precision even if that answer is based on an
inadequate model. A computer will not say “A population cannot be negative,” or “The
number of sick people cannot be greater than the total population.” The modeler must be
the one to verify the quality of a model. This paper and future Modeling Exercises in
Road Maps will enhance the reader’s modeling skills.

2
Powersim or Vensim can be used if the reader can make interpretations from the material in this paper.
8 D-4451-2

3. EXERCISE 1: URBAN DYNAMICS3

In this exercise the reader will build a simple model to explain growth and
stagnation in an urban area. Read the material in Section 3.1, then answer the questions in
Section 3.2. Section 3.3 further describes the system, and Section 3.4 gives instructions
to build a STELLA model to fit the description. The model should then be used to answer
the questions given in Section 3.5.

3.1 Background of the Urban System

A city’s life cycle is characterized by a period of economic growth followed by a
period of transition toward equilibrium. During the growth period, the city appears to be
economically healthy: business activity is expanding and unemployment is low. The
“Population” and the number of “Business Structures” grow quickly. During the
transition period, conditions become less desirable and pressures arise which impede
further growth. The “Construction” rate of “Business Structures” becomes smaller, and
so does the rate of “In Migration” into the city. In equilibrium, when the “Population” and
the number of “Business Structures” stop growing, the city suffers from problems such as
high unemployment.

3.2 Preliminary Exploration of the Urban System4

Please answer the following questions before continuing.
1. What is the problem the model will address? What is the reason for building the
model? What are the key variables in the system?

3
Note: The following exercise was based on an exercise prepared by John Sterman. It has been revised

for use in Road Maps. The model structure and description were adapted from the population-business

(POPBSN) model in Introduction to Urban Dynamics.

John Sterman, 1981. Principles Of Systems I, 15.852/872, Fall 1981, Assignment #3: Urban Modeling

Assignment (D-3296-1), System Dynamics Group, Massachusetts Institute of Technology.

Alfeld, L. E., and Graham, A. K., 1976. Introduction to Urban Dynamics. Cambridge, MA: Productivity

Press. The POPBSN model is developed in Chapter 5 of the book.

4
Suggested answers to this question are included in Section 5.

 D-4451-2 9

Sketch the expected behavior of all the key variables over an appropriate time horizon on
the blank graph provided.

Time

3.3 General Description of the Urban System

The availability of “Jobs” and promise of higher incomes are prime motivations for
“In Migration” into an urban area. People tend to move to areas where employment
opportunities are favorable. Television news, newspaper stories, advertisements, word-of­
mouth accounts, and industrial recruiting all tend to create perceptions of higher wages
and greater opportunities in areas where “Labor Availability” is low. On the other hand,
in areas where “Labor Availability” is high, job conditions are perceived to be relatively
unfavorable.5 People tend to migrate to cities that, at a given time, are perceived to offer
more job opportunities.
When people move into or out of an area, they add to or subtract from the number
of people in the area’s “Labor Force.” Businesses cannot ignore the “Labor Availability”
in their location and expansion decisions. Readily available labor allows businesses greater
10 D-4451-2

flexibility in choosing employees and shortens the time necessary to fill open positions.
Moreover, high “Labor Availability” tends to decrease wage competition for labor among
businesses.

3.4 Detailed Instructions for Building the Urban Growth Model

The urban growth model that the reader is asked to build in Section 3.5 should
focus on the dynamics of population and business growth within a fixed area of a city.
The model should contain two sectors: the “Business Structures” sector and the
“Population” sector.

3.4.1 The “Business Structures” Sector

The “Business Structures” sector consists of a level called “Business Structures”
with initial value of 1000. The number of “Business Structures” increases through a rate
of “Construction” and decreases through a rate of “Demolition.”
The rate of “Demolition” is determined by the number of existing “Business
Structures” and the “Average Structure Lifetime” of 50 years.
The rate of “Construction” can be modeled as a product of “Business Structures”
and several factors:6
? A “Construction Fraction” represents “Construction” as 2% of “Business
Structures” under normal conditions— when there is ample land, and labor availability
matches the available jobs. The “Construction Fraction” thus has a value of 0.02.
? A “Land Availability Multiplier” is a nonlinear function of the “Land Fraction
Occupied.” “Land Fraction Occupied” is defined as the number of “Business Structures”

5
“Labor Availability,” as it will be defined in the urban growth model, is the ratio of the “Labor Force” in
the urban area to the total number of “Jobs.” When the “Labor Force” is greater than the number of
“Jobs” (“Labor Availability” is greater than 1), not all the workers in the area can find “Jobs,” and the
area experiences unemployment. Thus, “Labor Availability” represents the concept of unemployment
from the point of view of the businesses.
 D-4451-2 11

multiplied by the average “Land per Structure” (1 acre per structure), and divided by the
total “Land Area” of 5000 acres. When almost all “Land Area” is available, the “Land
Availability Multiplier,” and thus “Construction,” is quite low because there are very few
businesses around to provide capital. As more of the “Land Area” is occupied, the effect
on “Construction” increases. Eventually, the “Land Area” is used up, and the “Land
Availability Multiplier” and “Construction” drop again, reaching zero when all the “Land
Area” is occupied. In equilibrium, “Land Availability Multiplier” is equal to 1.
? A “Labor Availability Multiplier” is a nonlinear function of “Labor
Availability.” “Labor Availability” is the ratio of “Labor Force” to “Jobs.” “Labor Force”
is the product of the “Population” and the “Labor Participation Fraction” of 0.35, or 35%.
“Jobs” is the product of “Business Structures” and 20 “Jobs per Structure.” When “Labor
Availability” is greater than 1 (“Labor Force” is greater than “Jobs”), businesses are more
likely to construct new “Business Structures.” When “Labor Availability” falls below 1,
the “Labor Availability Multiplier” and “Construction” fall. In equilibrium, “Labor
Availability Multiplier” is equal to 1.

3.4.2 The “Population” Sector

The “Population” sector consists of a level called “Population.” The initial value
of “Population” is 50000 people. “Population” increases through the rates of “Births” and
“In Migration,” and decreases through the rates of “Deaths” and “Out Migration.”
The rate of “Births” is a constant percentage of the “Population.” Approximately
1.5 people are born per 100 people per year.
The rate of “Deaths” is also a constant percentage of the “Population.” The
“Average Lifetime” of a person is approximately 67 years.
The rate of “Out Migration” equals 8% of the existing “Population.”
The rate of “In Migration” is also a product of “Population” and several factors:

6
In system dynamics, rates are often defined as a level multiplied by one or more factors. These factors
can be aggregated into the format of a “normal” value and one or more multipliers. The normal
represents the ratio of the rate to the level under standard conditions of equilibrium, while the multipliers
all vary around the value of 1 to include the effects of varying conditions on the system. It is considered
good practice if the multipliers are all equal to 1 in equilibrium. A later section of Road Maps will discuss
how this can be achieved.
12 D-4451-2

The “In Migration Normal” of 8% is the fraction of the “Population” that

immigrates into the city each year under normal conditions.
The “Job Attractiveness Multiplier” depends on “Labor Availability.” When there
are many available jobs (“Labor Availability” is less than 1 because “Labor Force” is
smaller than “Jobs”), the “Job Attractiveness Multiplier” increases to values greater than
1. People are inclined to move to the city. When available jobs are scarce (“Labor
Availability” is greater than 1 because “Labor Force” is greater than “Jobs”), the “Job
Attractiveness Multiplier” decreases to values between 0 and 1. People tend not to
immigrate to the city.

3.5 Exploration of the Urban Growth Model7

2. State the initial working hypothesis as to why “Business Structures” and “Population”
initially grow and eventually become stagnant. Use the description above for help with the
answer.

3. Using the instructions from Section 3.4, map out in STELLA the two interacting
sectors of the “Business Structures” and the “Population.”

4. Define the elements in the model and enter the model equations correctly into the
computer. Make sure to indicate clearly any numerical assumptions not given above (e.g.,
values assumed in nonlinear “graphical functions”). Try to make the model clear so that
others can understand its structure easily. Then simulate the model. Did the model
produce a behavior similar to that predicted earlier?

5. What causes the growth of “Population” and “Business Structures” of an urban area
during the early years of its development? Use the structure of the model to answer this
question. How does the answer relate to the answer to question 2?

7
Suggested answers to the following questions are included in Section 5.
 D-4451-2 13

6. What is the behavior of “Labor Availability” during the life cycle of an urban area?
What causes the behavior?

7. How does the finite “Land Area” come to limit growth? Will all the available “Land
Area” be filled up in equilibrium?
Use the model to show the effects of annexing surrounding land to increase the “Land
Area” to 7500 acres. Initialize the model in equilibrium.8 Use the STEP function in
STELLA to increase the “Land Area” in year 10. How does the behavior change? Will
all the “Land Area” be filled up in equilibrium?

8. Does the assumption of a fixed “Land Area” invalidate the model? Most cities can and
have expanded from their original areas. How does such expansion influence the results
given by the model? For example, what would be the likely consequences of expanding
the “Land Area” within which the city is allowed to grow? Simulate the model with a
sequence of such expansions, say every 20 years. (The reader might want to refer back to
the analysis in question 7.)

8
To initialize the model in equilibrium, first find the equilibrium values of the two levels, “Business
Structures” and “Population,” from the base run of the model. Then enter the equilibrium values as initial
values of the respective levels. The model should now remain in equilibrium during the entire simulation.
14 D-4451-2

4. EXERCISE 2: MENTAL SIMULATION

Mental simulation is a key tool of system dynamics. It is often tempting to run a
model immediately after building it on the computer. It is equally tempting to trust the
results from the first simulation run and assume that the modeler has adequately explored
the system. In order for system dynamics modeling to be a useful tool, the modeler must
be able to mentally simulate models. Mental simulation forces the modeler to thoroughly
understand the reasons behind the model’s behavior, and helps find problems of computer
models.
In the following exercise, the reader will use the model shown in Figure 1 to
perform mental simulation.

Rate A

Level 1 Level 2
Rate B

Time Delay

Figure 1: A simple model used for practicing mental simulation skills

 D-4451-2 15

The model in Figure 1 is analogous to the physical system shown in Figure 2.

Figure 2: The physical equivalent

In the physical system, the level of water in the containers corresponds to “Level
1” and “Level 2.” The pump corresponds to a constant “Rate A.” “Rate B” represents
the spigot on the first container that flows at a rate proportional to the level of fluid in the
upper container (“Level 1”).

The model is set up with the following conditions:

? “Level 1” is initially 0 gallons

? “Level 2” is initially 4 gallons
? The “Time Delay” is equal to 2 minutes
? “Rate A” is equal to 1 gallon per minute
? “Rate B,” in gallons per minute, is defined by:
16 D-4451-2

Level 1
Rate B =
Time Delay

4.1 Exploration of the Mental Simulation Model

In the following exercises the reader will mentally simulate the model under three
different conditions. Include as much information relating to the behavior of the model as
possible. Only simple calculations are necessary. Some things that the reader might want
to include are:

? initial and final value for “Level 1” and “Level 2”

? initial slope of “Level 1” and “Level 2”

? time constant of exponential behavior

? time-scale over which behavior occurs

 D-4451-2 17

1. Mentally simulate the model with the initial conditions described in the previous
section (“Level 1” = 0; “Level 2” = 4). Sketch the behavior of all model components on
the blank graphs. Include as much information as possible.

Level 1 and Level 2

4.00

0.00

- 4.00
0.00 3.00 6.00 9.00 12.00
Time

Rate A and Rate B

2.00

1.00

0.00
0.00 3.00 6.00 9.00 12.00
Time
18 D-4451-2

2. Mentally simulate the same scenario (“Level 1” = 0; “Level 2” = 4). This time, change
the value of “Time Delay” to 4.

Level 1 and Level 2

4.00

0.00

- 4.00
0.00 3.00 6.00 9.00 12.00
Time

Rate A and Rate B

2.00

1.00

0.00
0.00 3.00 6.00 9.00 12.00
Time
 D-4451-2 19

3. Now, assume that the “Time Delay” is equal to 2 again, but “Level 1” is initialized at 4
and “Level 2” at 0. Sketch what happens.

Level 1 and Level 2

4.00

0.00

- 4.00
0.00 3.00 6.00 9.00 12.00
Time

Rate A and Rate B

2.00

1.00

0.00
0.00 3.00 6.00 9.00 12.00
Time
20 D-4451-2

5. SOLUTION TO URBAN DYNAMICS EXERCISE

The following is one possible set of solutions to the questions posed in the exercise
on urban growth.

1. Problem Statement: The rapid early growth of “Population” and “Business

Structures” of an urban system ultimately gives way to a subsequent stagnant equilibrium
characterized by a tight, static job market. After the initial growth, the total number of job
opportunities per worker declines slowly, as “Business Structures” and “Population” move
toward equilibrium. The reason for building the urban growth model is to explain the
characteristic behavior of S-shaped growth that the system exhibits. The key variables of
the system are “Business Structures,” “Population,” and “Labor Availability.” The time
horizon over which the behavior occurs is approximately 60 to 80 years. Figure 3 shows a
graph of the expected behavior of the key variables. The sketch shows the transition to
equilibrium with high “Labor Availability” (unemployment) that was described in Section
3.1.

Population

Labor Availability

Business Structures

0 40 80
Years

Figure 3: System behavior

 D-4451-2 21

2. The “Construction” of “Business Structures” is encouraged when “Land Fraction

Occupied” is low, but becomes constrained as the “Land Area” fills up. People are
attracted to the city at first when the “Job Attractiveness Multiplier” is high. As
“Population” increases, “Labor Availability” increases to keep pace with business
expansion, thereby encouraging further development. The success of previous
investments also attracts more development. Eventually, the increased “Land Fraction
Occupied” decreases the “Land Availability Multiplier,” and thus causes “Construction” to
fall. The employment situation deteriorates because business “Construction” no longer
keeps pace with demand for jobs. The city stagnates with a job shortage and most of the
“Land Area” is occupied. The job shortage offsets the inherent attractiveness of the area
to potential migrants.

3. Figure 4 shows a STELLA model for the urban growth system. Detailed model
equations are included in Section 7.1 of the Appendix.
22 D-4451-2

Business Sector

Jobs
Jobs per Structure

Average Structure Lifetime

Construction Fraction Business Structures

Construction Demolition

~
Labor Availability Multiplier
Land per Structure

Labor Availability ~ Land Fraction Occupied

Land Availability Multiplier

Land Area

Population Sector

Labor Force
Labor Participation Fraction

~
Job Attractiveness Multiplier Population

In Migration Out Migration

In Migration Normal Out Migration Fraction

Births Deaths

Birth Fraction Average Lifetime

Figure 4: STELLA model of urban growth and stagnation

 D-4451-2 23

4. The model is a simple implementation of the description given in Section 3.4. The
“Business Structures” sector consists of one stock, “Business Structures,” which is filled
by “Construction” and drained by “Demolition.”
“Demolition” is defined by the familiar average lifetime formulation: the number
of existing “Business Structures” divided by the “Average Structure Lifetime.”
“Construction” is defined by the number of existing “Business Structures”
multiplied by a “Construction Fraction” and two multipliers, the “Land Availability
Multiplier” and the “Labor Availability Multiplier.”
The “Land Availability Multiplier” incorporates the effect of limited development
“Land Area” on the “Construction” rate. It is determined by the “Land Fraction
Occupied,” which is the amount of land occupied by existing “Business Structures”
divided by the total “Land Area” available for commercial development. For low values of
“Land Fraction Occupied,” increasing business development will lead to higher values for
the “Land Availability Multiplier.” The high values occur because of the more advanced
infrastructure associated with denser development. As the “Land Fraction Occupied”
increases further, more desirable land is used up and land prices begin to rise, so
“Construction” decreases. Eventually, when all the “Land Area” is used up and the “Land
Fraction Occupied” is 1, “Construction” must fall to zero. The graphical function for the
“Land Availability Multiplier” is shown in Figure 5.
24 D-4451-2

Figure 5: “Land Availability Multiplier” as a function of “Land Fraction

Occupied”

The “Labor Availability Multiplier” incorporates the effect of varying human

resources on the “Construction” of new “Business Structures.” It is determined by the
value of “Labor Availability,” the ratio of “Labor Force” to “Jobs.” When “Labor
Availability” is low, there is little incentive to construct new “Business Structures”
because employees are difficult to find. The “Labor Availability Multiplier” does not go
all the way to zero because some “Construction” will still occur based on the investors’
assumption that the creation of “Jobs” will cause “In Migration.” As “Labor Availability”
increases, so does the incentive to construct new “Business Structures.” Eventually, the
“Labor Force” exceeds the number of “Jobs” to a great enough extent that the effects of
high unemployment outweigh the advantages of high “Labor Availability,” so the “Labor
Availability Multiplier” stops increasing and levels off. The graphical function for the
“Labor Availability Multiplier” is shown in Figure 6.

Figure 6: “Labor Availability Multiplier” as a function of “Labor

Availability”

The “Population” sector contains the “Population” stock that is affected by

“Births,” “Deaths,” “In Migration,” and “Out Migration.” “Births” are equal to 1.5% of
the “Population” per year. “Deaths” are equal to the “Population” divided by the
 D-4451-2 25

“Average Lifetime” of 67 years. “Out Migration” is assumed to be a constant 8% of the

“Population” per year. “In Migration” is defined as the “Population” multiplied by the “In
Migration Normal” of 8% per year and a “Job Attractiveness Multiplier.”
The “Job Attractiveness Multiplier” shows the effect of the “Labor Availability” on
the number of people migrating into the urban area. When “Labor Availability” is low, the
job market is very favorable to the working part of the “Population.” The number of
“Jobs” exceeds the “Labor Force,” so good “Jobs” are relatively easy to come by.
Naturally, people are attracted to move into the urban area, so the “Job Attractiveness
Multiplier” is greater than 1. The curve is almost flat at the far left because it is assumed
that when the conditions are already extremely favorable, small changes one way or the
other will have little effect. As the “Labor Availability” increases, the employment
conditions begin to decrease, so the tendency to migrate into the area declines.
Eventually, when the conditions are extremely poor, the “Job Attractiveness Multiplier”
falls to nearly zero. At this point, there is almost no “In Migration.” The graphical
function of “Job Attractiveness Multiplier” is shown in Figure 7.

Figure 7: “Job Attractiveness Multiplier” as a function of “Labor Availability”

Figure 8 shows the base run of the urban growth model.

26 D-4451-2

1: Business Structures 2: Population 3: Labor Availability

1: 6000.00
2: 300000.00
3: 2
1.20

2
1
3
1
3
1: 3000.00
2: 150000.00
3: 1.00 3 3
2

2
1

1: 0.00
2: 0.00
3: 0.80
0.00 20.00 40.00 60.00 80.00
Years

Figure 8: Base run of the Urban Growth Model

5. In the first portion of the model simulation, the city is attractive to residents and to
people planning to move to the city because “Jobs” are readily available (“Labor
Availability” is less than 1). The increasing “Population” and the success of previous
investments demonstrate the potential for further growth and the profitability of
“Construction” of “Business Structures.” The development of the urban infrastructure
caused by business expansion permits further development to occur. The continued
availability of “Land Area” during the growth stage allows business “Construction” to
keep pace with the increasing “Population” so people continue to be attracted by the job
opportunities in the area. The hypothesis stated in the answer to question 2 correctly
identifies the reasons for the initial growth.

6. The quick initial increase in “Labor Availability” is due to the initial values of
“Population” and “Business Structures” that are “out of balance” with one another. After
the initial growth phase of “Business Structures” and “Population,” during which “Labor
Availability” declines slightly (in other words, job availability improves slightly), “Jobs”
become less available to the average job-seeker (“Labor Availability” starts increasing).
The lower availability of “Jobs” occurs because “Jobs” begin rising slower than
“Population” once the “Land Area” constraint slows business “Construction.” The
increase in “Labor Availability” (corresponding to a decline in job availability) continues
 D-4451-2 27

until a sufficiently large unemployment rate is generated to discourage further “In

Migration.” As noted earlier, in equilibrium, a job shortage must exist in the system in
order to offset the inherent attractiveness of the area.

7. The constraint of a fixed “Land Area” limits growth by making additional

“Construction” of “Business Structures” less and less economically justifiable. The model
reaches equilibrium when the “Land Availability Multiplier” equals 1 and thus reduces
“Construction” until it is equal to “Demolition.” Because “Construction” must be greater
than zero, the “Land Fraction Occupied” must be less than 1 in equilibrium. In the base
run, 90% of the “Land Area” is occupied in equilibrium.
Figure 9 shows the effect of annexing 50% additional “Land Area” to allow further
urban expansion. This is a common policy used to counter the effects of urban stagnation.
The model was initialized in equilibrium with the final equilibrium values from the base
run. In year 10, the available “Land Area” was raised from 5000 to 7500 acres. Figure 9
shows that expanding the “Land Area” available creates a new expansion of “Business
Structures” and “Population.” The expansion causes employment conditions to improve,
but the relief is only temporary and the same problem of high “Labor Availability” in
equilibrium exists. Again only 90% of the “Land Area” is occupied in equilibrium.

1: Business Structures 2: Population 3: Labor Availability

1: 8000.00
2: 500000.00
3: 1.20
1 2 1 2
1 2
3 3 3
3
1 2
1: 4000.00
2: 250000.00
3: 1.00

1: 0.00
2: 0.00
3: 0.80
0.00 20.00 40.00 60.00 80.00
Years

Figure 9: Equilibrium for 50% step increase in available “Land Area”

28 D-4451-2

8. Assuming a fixed “Land Area” simplifies real life. Cities do expand their geographical
boundaries, sometimes dramatically (consider Los Angeles and Mexico City). When total
“Land Area” available is expanded in the simple two-level model, “Business Structures”
and “Population” grow to fill the larger “Land Area.” However, the system still
equilibrates with high “Labor Availability” (i.e., unemployment). If a series of expansions
of the “Land Area” available is tested, each expansion will stimulate a further growth
spurt. Each growth spurt will lower “Labor Availability” in the short term, due to
increased business “Construction.” However, as each growth period gives way to a new
equilibrium, “Labor Availability” will rise again to its previous equilibrium value.
Therefore, expanding the city's geographic limits will lead to a temporary improvement in
the availability of “Jobs,” but the improvement will eventually be reversed as the new
“Land Area” is filled. The high unemployment equilibrium can only be avoided if the city
continues to grow indefinitely. Because no city can expand indefinitely, analyzing growth
within a fixed “Land Area” becomes a useful approach to studying the types of pressures
that can limit growth.
 D-4451-2 29

6. SOLUTION TO MENTAL SIMULATION EXERCISE

6.1 A Note on Time Constants

An understanding of time constants is useful in mental simulation of models. A

time constant is a number that relates a flow to the stock it affects. The definition of a
time constant is the reciprocal of the number which, multiplied by the level, defines the
rate. The model in Figure 11 shows one of the simplest examples of a time constant.

Stock

Flow

Time Constant

Figure 11: A simple model using a time constant

The rate equation in the model in Figure 11 is:

Stock
Flow =
Time Constant

The model contains one negative feedback loop. The “Time Constant” affects the
“Flow” that drains the “Stock.” The structure gives the following behavior for the
“Stock”:
30 D-4451-2

1: Stock
1.00
1

elapsed time
- Time Constant
Stock(t) = Stock(0) * e
0.50

1
0.00
0.00 1.00 2.00 3.00 4.00
Time Constants

Figure 12: Behavior of a single negative loop9

6.2 Mental Simulation Exercise Solution

1. Initial model run: Figures 13 and 14 show the behavior of the model. Both levels
converge asymptotically to an equilibrium valuethat can be found by setting the net rate at
each level equal to zero. Equilibrium occurs when “Rate A” equals “Rate B.” When
“Level 1”/“Time Constant” equals 1, the system will be in equilibrium, and the equilibrium
value for “Level 1” is 2. “Level 2” is affected by the same rates as “Level 1,” but in the
opposite direction. As a result, any change in “Level 1” is mirrored by “Level 2.”
Because “Level 1” increased by 2, “Level 2” decreased by 2 to a final value of 2.
The slope of the graph of a level is simply the net rate entering the level. At the
beginning of the simulation, “Level 1” is empty, so “Rate B” is 0. Therefore the only non­
zero rate is “Rate A” which equals 1. The initial slope of “Level 1” is +1 and the initial
slope of “Level 2” is –1.
The rest of the behavior can be determined using the “Time Constant” of the
system, which is 2. The shape of the curve can be determined accurately with the chart
and description outlined above. In Figure 13, the first three time constants are labeled.

9
Mathematically inclined readers may realize that the negative sign in the exponent causes asymptotic
growth. If the exponent is positive, growth is exponential.
 D-4451-2 31

1: Level 1 Initial Slope = –1 2: Level 2

4.00 1 time constant
2 time constants
2
3 time constants
2 2
1 1 2
1

1
0.00

Initial Slope = 1

-4.00
0.00 3.00 6.00 9.00 12.00
Time
Figure 13: First run behavior of “Level 1” and “Level 2”

1: Rate A 2: Rate B
2.00

1.00 1 1 1 2 1 2
2

0.00
0.00 3.00 6.00 9.00 12.00
Time

Figure 14: First run behavior of “Rate A” and “Rate B”

32 D-4451-2

2. Changing the “Time Delay”: Figures 15 and 16 show the behavior of the second run.
For the second run, the “Time Delay” was changed from 2 to 4. As a result, the
equilibrium value for “Level 1” moved from 2 to 4. In addition, the approach to
equilibrium was half the speed because the time constant was twice as large. As before,
the change in “Level 2” mirrored “Level 1,” so “Level 2” went from 4 to 0.

1: Level 1 2: Level 2
4.00
1
2 1

2
1 2
0.00

-4.00
0.00 3.00 6.00 9.00 12.00
Time

Figure 15: Second run behavior of “Level 1” and “Level 2”

 D-4451-2 33

1: Rate A 2: Rate B
2.00

1.00 1 1 1 1
2
2

0.00
0.00 3.00 6.00 9.00 12.00

Time
Figure 16: Second run behavior of “Rate A” and “Rate B”
34 D-4451-2

3. Changing initial conditions: Figures 17 and 18 show the behavior of the model when
the initial value of “Level 1” is changed from 0 to 4, and “Level 2” is initialized at 0. For
the third model run, the time constant and equilibrium value for “Level 1” are the same as
in the first run. However, because the initial value of “Level 1” is different, the model
takes a different path to equilibrium. The first noticeable changes are the initial slopes for
the levels. Because “Rate B” is not initially 0, but rather 2, the initial slopes of “Level 1”
and “Level 2” are –1 and +1 respectively (the reverse of the first run). Again, the behavior
of “Level 2” mirrors that of “Level 1,” so “Level 2” increases from 0 to 2.

1: Level 1 2: Level 2
4.00
1

1
1 1 2
2 2

0.00

-4.00
0.00 3.00 6.00 9.00 12.00
Time

Figure 17: Third run behavior of “Level 1” and “Level 2”

 D-4451-2 35

1: Rate A 2: Rate B
2.00

2
1.00 1 1 1 2 1 2

0.00
0.00 3.00 6.00 9.00 12.00
Time

Figure 18: Third run behavior of “Rate A” and “Rate B”

36 D-4451-2

7. APPENDIX

7.1 Equations for Urban Growth Model

Business Sector
Business_Structures(t) = Business_Structures(t - dt) +
(Construction - Demolition) * dt

INIT Business_Structures = 1000

DOCUMENT: Number of business structures in the city.

Units: structures

INFLOWS:

Construction = Business_Structures * Construction_Fraction *

Labor_Availability_Multiplier *
Land_Availability_Multiplier
DOCUMENT: Rate of construction of business structures. It is affected by the
number of already existing business structures, a normal construction fraction, and
the availability of land and labor.
Units: structures/year

OUTFLOWS:

Demolition = Business_Structures/Average_Structure_Lifetime

DOCUMENT: Rate of demolition of business structures.

Units: structures/year

Average_Structure_Lifetime = 50

DOCUMENT: Average lifetime of a business structure.

Units: years

Construction_Fraction= 0.02
DOCUMENT: Normal rate of construction of business structure per existing
business structure.
Units: 1/year

Jobs = Business_Structures * Jobs_per_Structure

DOCUMENT: Number of jobs provided by the existing business structures. It is
the product of the number of business structures and the average number of jobs
per structure.
Units: jobs

Jobs_per_Structure = 20

DOCUMENT: The number of jobs provided by each structure.

Units: jobs/structure

Labor_Availability = Labor_Force/Jobs
 D-4451-2 37

DOCUMENT: The ratio between the labor force and the number of available
jobs.
Units: people/job

Land_Area = 5000

DOCUMENT: Total land area available for commercial development.

Units: acres

Land_Fraction_Occupied = Business_Structures * Land_per_Structure/Land_Area

DOCUMENT: The fraction of the commercial land that has already been
developed.
Units: dimensionless

Land_per_Structure = 1

DOCUMENT: The amount of land required by each business structure.

Units: acres/structure

Labor_Availability_Multiplier = GRAPH(Labor_Availability)

(0.00, 0.05), (0.2, 0.105), (0.4, 0.225), (0.6, 0.36), (0.8, 0.54), (1.00, 0.84), (1.20,

1.24), (1.40, 2.36), (1.60, 3.34), (1.80, 3.86), (2.00, 4.00)

DOCUMENT: The multiplier shows the effect of Labor Availability on

Construction. When Labor Availability is greater than 1, businesses are more

likely to construct new structures.

Units: dimensionless

Land_Availability_Multiplier = GRAPH(Land_Fraction_Occupied)

(0.00, 1.00), (0.1, 2.30), (0.2, 2.98), (0.3, 3.34), (0.4, 3.48), (0.5, 3.50), (0.6,

3.44), (0.7, 3.12), (0.8, 2.30), (0.9, 1.00), (1, 0.00)

DOCUMENT: The multiplier shows the effect of availability of land on business

construction. When nearly all the land is available, Construction is slow because

there are very few businesses around to provide capital. As more of the Land Area

is occupied, the effect on Construction increases. As the available Land Area is

used up, Construction drops until it is zero at maximum capacity.

Units: dimensionless

Population Sector
Population(t) = Population(t-dt)+(In_Migration + Births - Out_Migration - Deaths)*dt
INIT Population = 50000
DOCUMENT: The number of people living in the urban area.
Units: people

INFLOWS:

In_Migration = Population * In_Migration_Normal *

Job_Attractiveness_Multiplier

38 D-4451-2

DOCUMENT: The number of people who move into the urban area each year. It
is affected by the current population, a normal fraction of in-migration, and the
availability of jobs.
units: people/year

Births = Population * Birth_Fraction

DOCUMENT: The number of people born in the area per year.

Units: people/year

OUTFLOWS:

Out_Migration = Population * Out_Migration_Fraction

DOCUMENT: The number of people who leave the urban area each year.

Units: people/year

Deaths = Population / Average_Lifetime

DOCUMENT: Number of people who die each year.

Units: people/year

Average_Lifetime = 66.7
DOCUMENT: The average lifetime of a person living in the urban area is
approximately 67 years.
Units: years

Birth_Fraction = 0.015

DOCUMENT: The fraction of the population that reproduces each year.

Units: 1/year

In_Migration_Normal = 0.08
DOCUMENT: The fraction of the population that immigrates each year under
normal conditions.
Units: 1/year

Labor_Force = Population * Labor_Participation_Fraction

DOCUMENT: The number of people who are eligible to work. It is a constant
fraction of the population.
Units: people

Labor_Participation_Fraction = 0.35
DOCUMENT: The fraction of the total population that is willing and able to
work.
Units: dimensionless

Out_Migration_Fraction = 0.08

DOCUMENT: The fraction of the population that emigrates each year.

Units: 1/year

 D-4451-2 39

Job_Attractiveness_Multiplier = GRAPH(Labor_Availability)
(0.00, 4.00), (0.2, 3.95), (0.4, 3.82), (0.6, 3.56), (0.8, 2.86), (1.00, 1.24), (1.20,
0.64), (1.40, 0.32), (1.60, 0.18), (1.80, 0.105), (2.00, 0.075)
DOCUMENT: The multiplier shows the effect of Labor Availability on
immigration. When there are many available jobs (Labor Availability is less than
1), people are inclined to move to the city. When there are not enough available
jobs (Labor Availability is greater than 1), people tend not to immigrate to the
urban area.
Units: dimensionless

7.2 Equations for Mental Simulation Model

Level_1(t) = Level_1(t - dt) + (Rate_A - Rate_B) * dt

INIT Level_1 = 4

Units: gallons

INFLOWS:

Rate_A = 1

Units: gallons/minute

OUTFLOWS:

Rate_B = Level_1/Time_Delay

Units: gallons/minute

Level_2(t) = Level_2(t - dt) + (Rate_B - Rate_A) * dt

INIT Level_2 = 0

Units: gallons

INFLOWS:

Rate_B = Level_1/Time_Delay

Units: gallons/minute

OUTFLOWS:

Rate_A = 1

Units: gallons/minute

Time_Delay = 2

Units: minutes

 40 D-4451-2

Vensim Examples:

Modeling Exercises

Section 2

Aaron Diamond & Nathaniel Choge

October 2001
JOBS PER
STRUCTURE jobs

CONSTRUCTION INITIAL BUSINESS

FRACTION STRUCTURES

Business
Structures demolition
construction

AVERAGE

STRUCTURE

LIFETIME

labor availability land availability

multiplier LAND PER
multiplier
land fraction STRUCTURE
occupied
LABOR
labor availability AVAILABILITY
LOOKUP LAND LAND AREA
JOBS PER PERSON AVAILABILITY
LOOKUP
job attractiveness labor force LABOR PARTICIPATION
multiplier FRACTION
JOB
ATTRACTIVENESS OUTMIGRATION

LOOKUP FRACTION

INMIGRATION inmigration Population outmigration

NORMAL
births deaths
INITIAL
BIRTH FRACTION AVERAGE LIFETIME
POPULATION

Figure 19: Vensim equivalent of Figure 4: model of urban growth and stagnation
 D-4451-2 41

Documentation for the Urban Growth Model:

(01) AVERAGE LIFETIME=66.7

Units: year
The average lifetime of a person living in the urban area is approximately 66.7
years.

(02) AVERAGE STRUCTURE LIFETIME=50

Units: year
Average lifetime of a business structure.

(03) BIRTH FRACTION=0.015

Units: 1/year
The fraction of the Population that reproduces each year.

(04) births=Population*BIRTH FRACTION

Units: people/year
The number of people born in the area per year.

(05) Business Structures= INTEG (construction-demolition, INITIAL BUSINESS

STRUCTURES)
Units: structures
Number of business structures in the city.

(06) construction=Business Structures*CONSTRUCTION FRACTION*labor

availability multiplier*land availability multiplier
Units: structures/year
Rate of construction of business structures. It is affected by
the number of already existing business structures, a normal
construction fraction, and the availability of land and labor.
42 D-4451-2

(07) CONSTRUCTION FRACTION=0.02

Units: 1/year
Normal rate of construction of business structure per existing business structure.

(08) deaths=Population/AVERAGE LIFETIME

Units: people/year
Number of people who die each year.

(09) demolition=Business Structures/AVERAGE STRUCTURE LIFETIME

Units: structures/year
Rate of demolition of business structures.

(10) FINAL TIME = 80

Units: year
The final time for the simulation.

(11) INITIAL BUSINESS STRUCTURES=1000

Units: structures

(12) INITIAL POPULATION=50000

Units: people

(13) INITIAL TIME = 0

Units: year
The initial time for the simulation.

(14) inmigration=Population*INMIGRATION NORMAL*job attractiveness multiplier

Units: people/year
The number of people who move into the urban area each year. It is affected by
the current Population, a normal fraction of in-migration, and the availability of
jobs.
 D-4451-2 43

(15) INMIGRATION NORMAL=0.08

Units: 1/year
The number of pepole who are eligible to work. It is a constant
fraction of the population.

(16) JOB ATTRACTIVENESS LOOKUP=([(0,0)(10,10)],(0,4),(0.2,3.95),(0.4,3.82),

(0.6,3.56),(0.8,2.86),(1,1.24),(1.2,0.64),(1.4,0.32),(1.6,0.18),(1.8,0.105),(2,0.075)
)
Units: dmnl

(17) job attractiveness multiplier=JOB ATTRACTIVENESS LOOKUP (labor

availability/JOBS PER PERSON)
Units: dmnl
The multiplier shows the effect of labor availability on
inmigration. When there are many available jobs (labor
availability is less than 1), people are inclined to move to the
city. When there are not enough available jobs (labor
availability is greater than 1), people tend not to immigrate to
the urban area.

(18) jobs=Business Structures*JOBS PER STRUCTURE

Units: jobs
Number of jobs provided by the existing structures. It is the
product of the number of business structures and the average
number of jobs per structure.

(19) JOBS PER PERSON=1

Units: people/jobs

(20) JOBS PER STRUCTURE=20

44 D-4451-2

Units: jobs/structures
The number of jobs provided by each structure.

(21) labor availability=labor force/jobs

Units: people/jobs
The ratio between the labor force and the number of available jobs.

(22) LABOR AVAILABILITY LOOKUP([(0,0)-(10,10)],(0,0.05), (0.2,0.105),

(0.4,0.225),(0.6,0.36),(0.8,0.54),(1,0.84),(1.2,1.24),(1.4,2.36),(1.6,3.34),(1.8,3.86
),(2,4))
Units: dmnl

(23) labor availability multiplier=LABOR AVAILABILITY LOOKUP (labor

availability/JOBS PER PERSON)
Units: dmnl
The multiplier shows the effect of labor availability on
Construction. When labor availability is greater than 1,
businesses are more likely to construct new structures.

(24) labor force=Population*LABOR PARTICIPATION FRACTION

Units: people
The number of people who are eligible to work. It is a constant
fraction of the population.

(25) LABOR PARTICIPATION FRACTION=0.35

Units: dmnl
The fraction of the total Population that is willing and able to
work.

(26) LAND AREA=5000

 D-4451-2 45

Units: acres
The total land area available for commercial development.

(27) LAND AVAILABILITY LOOKUP([(0,0)­

(10,10)],(0,1),(0.1,2.3),(0.2,2.98),(0.3,3.34),
(0.4,3.48),(0.5,3.5),(0.6,3.44),(0.7,3.12),(0.8,2.3),(0.9,1),(1,0))
Units: dmnl
The multiplier shows the effect of availability of land on business construction.
When nearly all the land is available, construction is slow because there are very
few businesses
around to provide capital. As more of the LAND AREA is occupied, the effect on
construction increases. As the available LAND AREA is used up, construction
drops until it is zero at maximum capacity.

(28) land availability multiplier=LAND AVAILABILITY LOOKUP (land fraction

occupied)
Units: dmnl

(29) land fraction occupied=Business Structures*LAND PER STRUCTURE/LAND

AREA
Units: dmnl
The fraction of the commercial land that has already been
developed.

(30) LAND PER STRUCTURE=1

Units: acres/structures
The amount of land required by each business structure.

(31) outmigration=Population*OUTMIGRATION FRACTION

Units: people/year
The number of people who leave the urban area each year.
46 D-4451-2

(32) OUTMIGRATION FRACTION=0.08

Units: 1/year
The fraction of the Population that emigrates each year.

(33) Population= INTEG (births+inmigration-deaths-outmigration, INITIAL

POPULATION)
Units: people
The number of people living in the urban area.

(34) SAVEPER = TIME STEP

Units: year
The frequency with which output is stored.

(35) TIME STEP = 0.0625

Units: year
The time step for the simulation.

Base Run of the Urban Growth Model

6,000 structures 2 2 2
300,000 people 2
1 1 1 1
1.2 people/jobs 2 1 3 3 3

1 3
3 2 3
3 3 3
2 1
0 structures 2 1
1
0 people 12
0.8 people/jobs
0 8 16 24 32 40 48 56 64 72 80
Time (year)
Business Structures : 1 1 1 1 1 1 1 structures
Population : 2 2 2 2 2 2 2 2 2 people
labor availability : 3 3 3 3 3 3 3 people/jobs
Figure 20: Vensim equivalent of Figure 8: Base run of the Urban Growth Model
 D-4451-2 47

Model Run with STEP INCREASE in LAND AREA

8,000 structures
500,000 people 12 12 12 12 1 2 1
12
1.2 people/jobs 3
3
3 3 3 3 3
12
12
3

0 structures
0 people
0.8 people/jobs
0 20 40 60 80 100
Time (Year)
Business Structures : 50% increase 1 1 1 1 1 structures
Population : 50% increase 2 2 2 2 2 2 people
Labor Availability : 50% increase 3 3 3 3 3 people/jobs

Figure 21: Vensim correction of Figure 9: Equilibrium for 50% step increase in available
“LAND AREA”

INITIAL LEVEL 2
TIME DELAY
Level 2

rate a rate b

Level 1

INITIAL LEVEL 1

Figure 22: Vensim Equivalent of Figure 1: A simple model used for practicing mental
simulation skills.
48 D-4451-2

Documentation for Mental Simulation Model:

(01) FINAL TIME = 12

Units: minutes
The final time for the simulation.

(02) INITIAL LEVEL 1=4

Units: gallons

(03) INITIAL LEVEL 2=0

Units: gallons

(04) INITIAL TIME = 0

Units: minutes
The initial time for the simulation.

(05) Level 1= INTEG (rate A-rate B,INITIAL LEVEL 1)

Units: gallons

(06) Level 2= INTEG (+rate B-rate A,INITIAL LEVEL 2)

Units: gallons

(07) rate a=1

Units: gallons/minutes

(08) rate b=Level 1/TIME DELAY

Units: gallons/minutes

(09) SAVEPER = TIME STEP

 D-4451-2 49

Units: minutes

The frequency with which output is stored.

(10) TIME DELAY=2

Units: minutes

(11) TIME STEP = 0.0625

Units: minutes
The time step for the simulation.

First Run Behavior of Levels

4 gallons 2
2
2 2
1 12 12 12 12 12 12 1
1
1
1
0 gallons

-4 gallons
0 2 4 6 8 10 12
Time (minutes)

Level 1 : 1 1 1 1 1 1 1 gallons
Level 2 : 2 2 2 2 2 2 2 gallons

Figure 23: Vensim equivalent of Figure 13: First run behavior of “Level 1 and Level 2”
(Level 1 is initially 0 and level 2 is initially 4 for the first and second runs)
50 D-4451-2

First Run Behavior of rates

2 gallons/minutes

1 gallons/minutes
1 1 1 1 1 1
2 2 2 1
2 2
2
2
2
0 gallons/minutes 2
0 2 4 6 8 10 12
Time (minutes)

rate a : 1 1 1 1 1 1 gallons/minutes
rate b : 2 2 2 2 2 gallons/minutes
Figure 24: Vensim equivalent of Figure 14: First run behavior of “rate a” and “rate b”

Second Run Behavior of Levels

4 gallons 2 1 1 1 1
1 1
2 1
1
12
2
1 2 2
0 gallons 2 2 2
1 2

-4 gallons
0 2 4 6 8 10 12
Time (minutes)

Level 1 : 1 1 1 1 1 1 1 gallons
Level 2 : 2 2 2 2 2 2 2 gallons
Figure 25: Vensim equivalent of Figure 15: Second run behavior of “Level 1” and “Level
2”
 D-4451-2 51

Second Run Behavior of rates

2 gallons/minutes

1 gallons/minutes
1 1 1 1 1 1 1 1
2
1
2
1
2 1
2 2
2
2
2
0 gallons/minutes 2
0 2 4 6 8 10 12
Time (minutes)

rate a : 1 1 1 1 1 1 gallons/minutes
rate b : 2 2 2 2 2 gallons/minutes

Figure 26: Vensim equivalent of Figure 16: Second run behavior of “rate a” and “rate b”

Third Run Behavior of Levels

1
4 gallons
1
1 1 1 12 12 12 12 12 1
2 2
2
2
0 gallons 2

-4 gallons
0 2 4 6 8 10 12
Time (minutes)

Level 1 : 1 1 1 1 1 1 1 gallons
Level 2 : 2 2 2 2 2 2 2 gallons
Figure 27: Vensim equivalent of Figure 17: Third run behavior of “Level 1” and “Level 2”
52 D-4451-2

Third Run Behavior of rates

2 gallons/minutes
2

2
1 gallons/minutes 2
2 21 2
1 1 1 12 2 1 2 1

0 gallons/minutes
0 2 4 6 8 10 12
Time (minutes)

rate a : 1 1 1 1 1 1 gallons/minutes
rate b : 2 2 2 2 2 gallons/minutes

Figure 28: Vensim equivalent of Figure 18: Third run behavior of “rate a” and “rate b”