Table of Contents
MacroLab
Model-Facilitated Learning of Essential Macroeconomics
David Wheat
PhD Candidate in System Dynamics, University of Bergen
Adjunct Professor, Virginia Western Community College
President, Wheat Resources Inc.
P.O. Box 19234
Roanoke, Virginia 24019
540-966-5939
540-966-5167 (fax)
dwheat@wheatresources.com
Abstract
MacroLab is a series of interactive learning environments (ILEs) for introductory
macroeconomics. It is currently used in a distance learning course taught by the
author. The underlying system dynamics structure consists of a national
accounting model of aggregate income and spending, linked with behavioral
models of consumption, investment, production and pricing, labor and capital
utilization, and fiscal and monetary policy. After a brief review of the problem
that motivated the development of MacroLab, this paper summarizes the rationale
for a solution based on system dynamics and model-facilitated learning principles,
highlights key features of the ILE series, and outlines a future research agenda.
Workshop participants will gain economic understanding by “test driving” some
of the ILEs. In addition, they will gain ILE development skills by experiencing
the step-by-step content delivery process, reviewing the corresponding design
process, and discussing analogous steps for ILEs in their own fields.
keywords: experiment, interactive learning environment, ILE, macroeconomics,
management flight simulator, learning lab, microworld, teaching.
1. The Problem
There is reason for concern over the efficacy of economics instruction. Experiments by
Walstad and Allgood (1999) indicate that the value added by economics courses is “minimal.”
Wood and Doyle (2002) report similar findings in a large economic literacy survey, concluding
that college and university economics educators should be only “modestly” encouraged by the
results.
With respect to macroeconomics in particular, Cohn et al. (2001) find no significant
difference in learning gain by students who received traditional instruction—graphical
comparative statics—on a Keynesian concept, compared to those who received only verbal
instruction. In a similar experiment involving a monetary policy topic, students receiving
graphical instruction actually scored /ower than those receiving verbal instruction alone.
My own students find it difficult to see how the proliferating textbook graphs relate to
one another, and how the variables on different graphs interact to influence economic system
performance. Indeed, interaction—inherently multiplicative—is de-emphasized by the graphical
approach, with its emphasis on stacking and adding the components of aggregate spending. The
elusive dynamics of interaction is almost certainly missed by students just struggling to keep
track of the ceteris paribus assumptions behind each new graph.
Boucher (1995) suggests the “...possibility that by over-concentration on comparative
statics, the [economics] profession may be omitting or playing down the key importance of the
dynamic aspects of most economic problems and issues.”
The routine content delivery technique may also impede learning. Becker and Watts
(1998) found that about eighty percent of economics classroom contact time was spent
lecturing—‘chalk and talk”—whether the courses were introductory or advanced. In contrast,
the use of techniques to get the students actively engaged in constructing knowledge was
minimal. Even in statistics and econometrics courses, just twenty-two percent of the available
time was spent in computer labs. Of course, heavy reliance on the lecture style is not prima facie
evidence of ineffective instruction. Nevertheless, when coupled with other research that is only
“modestly” encouraging to economics educators, such findings force the inevitable questions:
What might explain the apparent weakness in methods, and what solution is available?
2. Rationale for a Solution Based on System Dynamics and Model-Facilitated Learning
One hypothesis is that the traditional macroeconomics instructional method—graphical
comparative statics presented in a lecture format—provides a weakly structured learning pattern
for most economics students. Forty years ago, Bruner (1963) concluded that “...the most basic
thing that can be said about human memory...is that unless detail is placed into a structured
pattern, it is rapidly forgotten.”
Of course, the comparative statics model is very simple, compared to a complex
economic system characterized by reinforcing and counteracting feedback effects that are often
non-linear. If there is a weakness in the model, however, it is not because of simplicity. All
models— graphical, econometric, or simulation—are simplifications of reality and, as such, are
“wrong.” But (continuing to paraphrase) “...some models are useful.”"
Comparative statics fails the usefulness test if its purpose is to explain the behavior of an
economy during a transition from one equilibrium condition (or stable growth path) to another.
Apologists would argue that is not its purpose. However, the essential macroeconomics for most
adults is virtually synonymous with media treatment of such transitions, and public policy debate
is dominated by such topics. It is a purpose that needs to be served, and comparative statics fails
to do so.
' The phrase has become a truism in system dynamics circles, yet its origin is obscure, bordering on professional
folklore. Richmond (2000) attributes it to quality expert, W. E. Deming. More to the point, however, is that the
validity —and, therefore, the usefulness--of any model depends on its purpose, a principle emphasized by Forrester
(1961) four decades ago.
In Macroeconomic Essentials: Understanding Economics in the News, Kennedy (2000)
leads off with this rhetorical question:
What do we want our students to be able to do upon completing
their macroeconomics course—manipulate a 45°-line diagram and
derive the multiplier, or interpret and evaluate media commentary
on the macroeconomy? I am not alone in believing that for many
students the latter is preferable, and that in terms of meeting this
latter goal there is a high opportunity cost associated with
pressing students to learn algebraic and graphical derivations.
His textbook is a lucid exposition of fundamental macroeconomic principles, with
hundreds of media clips that serve to test student understanding and demonstrate the relevance
and utility of course content. The book contains only a few graphs, but it is nevertheless
rigorous in the critical thinking demands placed on student readers. What is missing, however, is
Bruner’s “structured pattern,” or, in Forrester’s words (1994): “...the framework into which facts
can be placed [so that] learning becomes more relevant and meaningful.”
I stand with those who believe that system dynamics provides such a framework, but that
is a testable hypothesis.” One research frontier is the development of methods for assessing the
instructional value added by interactive learning environments (ILEs) based on system dynamics
models. A prerequisite for rigorous evaluation, however, is a consensus on the principles of
effective ILE development and implementation.
Model Facilitated Learning. In their development of model-facilitated learning (MFL)
guidelines, Spector and Davidsen (1998, 2000) and Milrad, Spector, and Davidsen (2003)
delineate three stages of encounter between a learner and an ILE. At each stage, they urge a
unique instructional approach and expect a different learner development pattern.
The first stage promotes problem orientation. Learners confront a relatively simple
problem in the subject domain and must offer a solution. Next, the inquiry and exploration stage
makes the underlying model more transparent (Alessi 2000), and learners literally begin to see
the system structure, albeit in a very simplified form such as condensed causal loop or stock/flow
diagrams. This stage also includes “hypothesis formulation and experimentation” that challenge
users to articulate in words and diagrams their understanding of the relationship between system
structure and behavior. The final stage in the MFL process is the policy development stage,
where “learners are immersed in the full complex system and asked to develop rules and
heuristics to guide decision making in order to create stability or avoid undesirable situations.”
(Milrad, Spector and Davidsen, 2003). A common denominator for all stages is the principle of
graduated complexity that suggests a developmental sequence of learner challenges.
With only slight modification, these MFL principles frame the design of the MacroLab
user interface and the learning activities described in the third section of this paper. The final
product is a set of eighteen interactive learning units that comprise a system for learning
introductory macroeconomics.’
Underlying System Dynamics Model. Much of the structure of the model (that “facilitates
the learning”) in MacroLab builds on the work of those who pioneered the system dynamics
2 near .
~ See discussion in the research agenda section.
MacroLab is currently utilized in the author’s macroeconomics distance learning course for students at Virginia
Western Community College in Roanoke, Virginia. The series could also be used for self-paced learning.
approach to the study of macroeconomics, some of whom remain at the frontier of that research
initiative.
In addition, the model makes use of a traditional macroeconomics concept—circular flow
of income and spending—recast in stock and flow terms. Figure | is a simplified diagram of the
current version of the model, which joins “demand side” and “supply side” sectors of a national
economy.” Most link-forming arrows have been omitted to improve clarity. Arrows shown are
only those that are in the main reinforcing feedback loop for economic growth (or decline). The
diamond-shaped icons contain sub-models of various decision processes, within which most of
the counteracting feedback loops operate.
Figure 1. Simplified stock-and-flow diagram of the complete MacroLab model.
The demand sector—top half of Figure 1 —is a system dynamics model of the traditional
textbook “circular flow” diagram, where the material stocks are trillions of U.S. dollars, and the
flow of funds is driven by decisions that affect the timing and distribution of income, private
sector spending and saving decisions, and public sector taxation and spending decisions. The
bottom half of the figure displays the supply sector, and includes the forces driving production
* Any long list runs the risk of inadvertent omissions, but J. Forrester (1961, 1968, 1980, 1994), N. Forrester (1973,
1982), N. Mass (1975, 1980), J. Sterman (2000), and M. Radzicki (1988, 1993) certainly merit short-list
acknowledgement. I am deeply indebted to their pioneering efforts, as well as to the guidance provided by E.
Moxnes and P. Davidsen at the University of Bergen. Of course, they are in no way responsible for my
interpretations or adaptations, and weaknesses in the MacroLab model are solely my responsibility.
> This version models a closed economy, but an international sector will be added during the fall semester, 2003.
decisions, including the stocks of labor and capital and their associated productivity.° The two
sectors are linked by information feedback. Production information and prices/costs update
income payments and distributions. Capital decisions determine business saving and investment.
Real sales are equivalent to nominal aggregate spending divided by the price index.
Students using the MacroLab series do not encounter the model in Figure 1 until very late
in the macroeconomics course. After opening a MacroLab learning unit file, the users are quickly
led to an interface page that contains eighteen buttons, one for each learning unit (Figure 2).
[a i tnoavaton ] [bat 7 the Gerard Peare |] [aes he Bopp Pere fl
[ae a ease eae |] [Dates spay & beara | [pete eet Pater Poe
[bat s: credor Faw afieame & Spending | [Uni or Eewomic Growth & Ructators [Wn 151 Fiscal Polen on 2
[Wats consumption te crear Row | [nt 10s Labor utzon a coats | [Dnt isi monetary Polen Para
ee ee er |
[at 6 Government the Great Row] [Unk 12s Money Supp @ terest Rates) [lw 46s The 6 Pte |
Figure 2. Learning Unit Selection Panel
The entire series consists of eighteen ILEs that are stand-alone, model-facilitated learning
units. Together, however, they constitute a learning system. After the initial basic model is
introduced in Unit 2, subsequent learning units add dynamic complexity to that model without
changing its structural fundamentals. During the semester, the complete model is revealed and
becomes operational in stages. The design objective is for students to see the basic model
gaining complexity without losing sight of its fundamental structure.
Each of the learning units is a separate Stella file’. Clicking on any of the
buttons—except the red one—opens a text box that provides a brief summary of the selected
learning unit and describes the location of the file on the user’s computer. A click of the red
button (“Unit 2: Basic Model” in Figure 2) takes users to the interface page that activates the
instructional process for the selected unit.*
Among other things the user finds in Unit 2, there is the basic model of the entire
MacroLab series (Figure 3). It is the foundation structure that supports all subsequent
enhancements in later learning units. As in the complete model, the top half of the basic model
contains the demand decisions, while the bottom half contains the supply decisions. Decisions
© The fiscal policy and monetary policy decision points are located somewhat arbitrarily in Figure 1. Also, only the
forces contributing to the main reinforcing feedback loop (production>income>spending>sales>production) have
visible links. Other links were deleted for diagram clarity. The decision diamonds contain the negative feedback
loops that counteract the main reinforcing loop.
2 High Performance Systems (www.hps-inc.com) offers a free demo version of the STELLA simulation software.
8 Unit 1 introduces the MacroLab interface and the STELLA software tools, as well as fundamental system
dynamics concepts (stocks, flows, and feedback). It, too, is based on MFL principles but will not be discussed here.
about the timing and distribution of income payments link the production flow in the supply
sector with the income flow in the demand sector.
Demand
All Other
Bank Accounts
Income
Decisions
7 Business
# Inventories
9 production FRY sales nal
7
7
2
== Figure 3. Basic Model
Supply
Decisions
The visible loop in Figure 3 is a reinforcing loop showing that the supply side and the
demand side drive each other, and that both Say’s Law and Keynes’ Critique are incomplete
when considered in isolation.’
Inside the decision diamonds are sub-models of counteracting loops that limit the
ultimate strength of the reinforcing, economic growth loop. The degree to which that growth
loop is moderated depends on the adjustment times for the various counteracting feedback
processes. The embedded counteracting loops, along with other detailed aspects of Unit 2, are
discussed in the next section, which illustrates MacroLab’s model-facilitated learning process.
It is important to appreciate the fundamental relationship between the complete model in
Figure 1 and the basic model in Figure 3. Small pictures of both diagrams are combined in
Figure 4, and the similarity should be evident, despite differences in complexity.
On the left of Figure 4, the basic model is reproduced, while the complete model is on the
right. The circular flow of income and spending is visible in the top half (“demand side”) of
both diagrams, while the lower halves contain the supply decisions that drive production.
2 Keynes asserted that the macroeconomic views of Say and Ricardo could be summarized as “supply creates its
own demand.” He said they had it backwards. Using the basic model to distinguish the two theories is discussed in
the research agenda section.
Figure 4. Basic Model (left) vs. Complete Model (right)
3. Model-Faciltated Learning in MacroLab
This section illustrates how MFL principles have been adapted and implemented in
MacroLab, using excerpts from Unit 2. Unless indicated otherwise, the Unit 2 excerpts typify
the layout in each learning unit, which is organized into three main pages:
+ Home Page contains information about the purpose of the unit, a diagram of
the current model, and buttons that reveal the “story” behind the model.
+ MacroLab is the page containing the simulation control panel and the
learning activities and assignments.
+ Suggested Solutions is a reference page where users can check their work.
Home Page. Recall that the first MFL principle discussed in section two was “problem
orientation,” an initial exercise to acquaint users with issues they will confront in the learning
activities. And, indeed, the first simulation activity of Unit 2 provides such an exercise.
However, that exercise is in the context of the fundamental challenge for the entire
course—understanding economic fluctuations in the midst of long-run economic growth—a
challenge already introduced in a Unit | activity.
Therefore, the home page of Unit 2 (Figure 5) begins with a reminder of what should
have been learned in the previous unit, with the implication that the learning objectives are
cumulative, and that a review may be appropriate before proceeding.”
10 E.g., behavior of GDP and unemployment, fundamental system dynamics concept (stocks, flows, and feedback),
distinctions between mental models and computer models, and guided practice in the use of the software tools
necessary to operate the ILE.
Unit 2: Basic Model
The first unit acquainted you with the MacroLab interface by exploring
trends in GDP and unemployment over the past fifteen years. Also, Learning Objectives
you were introduced to the fundamental concepts of stocks, flows, for this Unit
feedback, and equilibrium, as well as to the basic procedure for
drawing causal loop diagrams.
In addition, you learned some of the ways to navigate the Stella
software environment, and use the tools to make things happen. That
is the essence of model-facilitated learning: making things happen,
observing the results, drafting hypotheses about why the underlying
system behaves at it does, and then "making things happen" again to
test the hypotheses under different conditions.
‘The primary purpose of this unit is to enable you to develop a
preliminary understanding of the basic model that is used in this course,
illustrated by the diagram at right. Click the button at top-right to see
the specific learning objectives. [oom out
oom" button to get a close look at the Zoom in for @ closer look at this small
diagram. Keeping t mind will be helpful as we Picture of the model in this unit,
develop more complex vari theme throughout the
‘course: liek the "story" buttons to learn more
Before proceeding, click the
i
Then go to the MacroLab to run the
The story buttons will present the diagram as a set of building blocks, simulations.
and reveal how the pieces fit together and work as a system. Read the
demand and supply stories prior to the "complete" story.
Figure 5. Home Page of a Typical MacroLab Learning Unit
The home page text also clarifies the primary goal of Unit 2: to enable the user to develop
a preliminary understanding of the basic model, including its reinforcing and counteracting
feedback loops. The goal is operationalized by a set of specific learning objectives that appear
when users click the button above the model diagram. Successful completion of the activities in
Unit 2 should enable students to:
1. Use a few more Stella tools effectively, and demonstrate a better grasp
of concepts introduced in the previous unit.
2. Write a brief summary of the demand and supply sides of the model,
and how they fit together.
3. Identify where, in principle, "GDP" fits in the model, and explain how
sales and inventory data can be used to estimate production.
4. Sketch a causal loop diagram that expresses an understanding of the
basic model in action, and explain that diagram.
5. Understand news media stories about the delayed effects of changes in
total (i.e., aggregate) spending in the economy.
Prior to arriving in Unit 2, students would have already encountered the mental model
concept in a Unit | activity and in supplementary reading materials. As a result, achieving
“understanding” in the course would have already been expressed in terms of renovation of the
students’ mental models.'' Thus, the first true learning experience in each unit is to get
acquainted with the “current” model in the MacroLab series (i.e., the model that is driving the
current learning unit), and that is accomplished in two steps.
First, on the Unit 2 home page, the user is encouraged to “zoom in” and get a close look
at the diagram of the model on that page. The home page diagram always contains the essential
stock and flow structure of the current model, but it is simplified by omitting links that would
add more clutter than clarity. It is, in effect, a type of influence diagram for communicating
primary relationships in the model, rather than a detailed stock-and-flow diagram.
After exploring the total picture, users are directed to the three “story” buttons in Unit
2. The “story” function in STELLA permits revealing portions of the model in a logical
sequence, accompanied by annotations." In Unit 2, the demand-side and supply-side stories
provide different perspectives on how an economic system functions, and the “complete” story
ties together the two perspectives in a loop-closing fashion."*
Clicking the “Demand Story” button reveals the portion of the model in Figure 6, a few
elements at a time. Each time the spacebar is pressed, the diagram takes shape, accompanied by
explanatory text.
Demand
Decisions
1 Business
Bank Accounts Bank
L_} L_}
Figure 6. Demand Side of Basic Model
Demand decisions are made by the spending agents in the model: consumers, businesses,
governments, and international customers. In the basic version of the model in Unit 2, the only
influence on demand decisions is information about the income received by the spending agents
"An instructor ass ‘ssing student progress in mental model renovation necessarily has some goal in mind, toward
which the students are hopefully moving. Most likely, the instructor’s own mental model provides both the implicit
target for the students, as well as the conceptual framework for instruction in the course. Instructors are expected to
teach what they know. In turn, society hopes that what instructors know is accurate and important, and that students”
mental models are being effectively renovated.
"? Most learning units have only one story button.
3 \ powerful feature of the story function is the capability to run a simulation using only the portion of the model
revealed at particular stages in the story. In Unit 1, for example, a story is used to reveal the stock adjustment
structure responsible for counteracting feedback. Prior to closing the feedback loop, however, a simulation run
reveals that the “incomplete” model generates linear growth when the goal of the system is shocked by a step
increase. Changing the adjustment time parameter merely affects the slope of the growth pattern. Only when the
loop is closed—on the next page of the story—does goal-seeking behavior emerge. That is an important revelation
for students who assume it is the delay parameter itself that creates the counteracting behavior. Running two
simulations within the story demonstrates that it is the feedback loop that generates the goal-seeking behavior.
' See the research agenda section for a discussion of issues raised by a biased instructional perspective (i.e., one
that fails to give a balanced treatment to both the demand- and supply-side sectors of the economy).
(including taxes received by government). Income is a (supply-side driven) flow of funds from
Business Bank Accounts to All Other Bank Accounts.
In this model, the essence of demand decision making is continuous revision and
implementation of spending goals in light of changes in income.’ Clicking on the demand
decisions diamond reveals a sub-model of the information stock adjustment process that has a
counteracting feedback effect on the positive link between income and spending. Demand
decisions are carried out by spending, which causes a flow of funds from "other bank accounts"
to "business bank accounts."
When users finish the story about the demand sector, they return to the home page and
click the “Supply Story” button, which reveals the portion of the model in Figure 7, a few
elements at a time.
Business
Inventories
© production sales 0
Figure 7. Supply Side of Basic Model
Within the model, supply decisions are made by producing agents (i.e., businesses). In
the basic model, information about the sales rate is the only factor that influences decisions about
how much to supply. Other influence factors (e.g., inventory levels, availability of labor and
capital, price and cost) are added as the model takes shape in later units. The essence of supply
decision making is the continuous revision and implementation of production goals. (Clicking
on the supply decision diamond reveals the sales information stock-adjustment process.)
Ultimately, the supply decisions affect the production rate of final goods and services.
Upon completion of the supply sector story, users return to the home page and click the
“Complete Story” button. Initially, the demand side of the model re-appears, along with a brief
summary of its previous story. Pressing the spacebar causes the supply side and its summary
story to re-appear. Then the connection between the two sides is accomplished by two steps that
create a closed loop. See Figure 8.
5 In later units, other influences on spending are added to the model.
firtg,
f s
es \
Step 1 1 a.
toclose " aN Step 2
the loop ae id a‘ to close
J | the loop
an
Paiet |
7 ma }
2. production o- J | re sales 9
® a
supply”
Decisions
Figure 8. Closing the Loop between Demand Side and Supply Side of Basic Model
The first step involves linking production and income. Income decisions are made by the
owners of production (or the producing agents on their behalf). The essence of income decision
making is the continuous revision and implementation of income payment goals. In this model,
that process is driven by information about current production rates. As before, the information
stock adjustment process is revealed by clicking on the decision diamond. Production results
from the combined services of the factors of production (labor, capital, raw materials, energy,
etc.). Payment for those services is factor income. When factor income is combined with profits
of the owners of the production, the total is national income
The second step reflects the fact that the spending flow and the sales flow are two sides
of the same coin. Although they represent the same real-world activity, there is a measurement
unit difference in the model. Spending is a flow of dollars. Sales, on the other hand, is a flow of
real goods and services. In this basic model, the sales rate is identically equal to the spending
rate. In later units when the model includes a price sector, sales will be equal to spending
divided by a price index."
Closing the loop confirms the mutual dependency of the demand and supply sides of the
model, and illustrates the futility of a “chicken or egg” search for primal causes. When the entire
MacroLab series is used as a learning tool, the recommended approach is to follow the numbered
sequence shown in Figure 2. Of course, that results in learning the “demand side” of the model
prior the “supply side.” That merely reflects a judgment about graduated complexity, however,
and has nothing to do with economic philosophy.'’ The demand-side units (3-7) are easier to
learn, and they correlate with the early material in standard textbook arrangements.
Nevertheless, if an instructor (or a self-paced learner) preferred to begin with the supply-side
units (8-13), there would be no problem since they can stand alone pedagogically.
MacroLab Control Panel Page. A click of the “MacroLab” button takes users to the page
containing the simulation controls, graphical outputs, and the associated activities and
assignments. The typical control panel contains another small picture of the current model,
16 . . er . . ape
Also, the conversion of real production to nominal income (Figure 8) requires multiplying by a price index.
"7 See the research agenda section for a related issue.
augmented by “on/off” switches and other parameter controls (e.g., sliders, dials, or input lists)
that permit testing the model under different assumptions. In addition to buttons containing
instructions and assignments, the control panel also includes a stacked set of graphs for
displaying the simulation results. Figure 9 displays the control panel page for Unit 2.
The dials in Figure 9 permit changes in the adjustment times for each information stock.
The spending shock switch causes a small (e.g., 2%) but sharp increase in spending in year 2,
only to be followed in year 4 by an equally sharp decrease (perhaps analogous to a two-year
period when strong demand abroad increased net exports). When the feedback switch is OFF,
any variation in spending is entirely due to the spending shock. When feedback is ON, then
spending will also be affected by changes in income. The random noise (i.e., disturbance) in the
production rate causes fluctuations that might mimic equipment problems, work stoppages,
supply bottlenecks, etc., in a real economy.
MacroLab
eee Unit 2: Basic Model
adj. delay
instructions | | Rexdthe
(ET
Sim [sim [Sim
aE Te | istgamene 22 |
Sim | sim Tee
| |e Assignment 2.3 Solutions
et
income —_ :
, ead 2 ele
adj. delays aE y 8
Bl 6
Run
Il
production om
adj. delay oli
00 abo ab abo abo 1000
sl Years
id Geaph 1. Spending and Sales ( trillon/yea)
Adlick leaf to turn page
Figure 9. MacroLab Control Panel for Unit 2
Sim A implements the “problem orientation” principle of model-facilitated learning by
requiring users to recall and use concepts introduced in Unit | (stocks, flows, equilibrium, and
GDP) when the model runs without any test inputs. Stocks and flows have to be identified, and
equilibrium has to be explained in stock/flow terms. Moreover, users have to associate GDP
with the production rate in the model, and recognize the essential equality of production, income
and spending/sales in equilibrium. Finally, the user must use a stock/flow understanding of
disequilibrium to explain that the sales flow plus changes in the inventory stock will equal the
production flow.
The remaining simulation activity and assignment buttons involve use of the spending
shock and random noise switches, first with feedback OFF and later with feedback ON. In
addition, the effects of changes in delay times (information stock adjustment times) are
evaluated. Through a series of questions, simulation runs, graphical results, the user is led to
conclusions about direction of causality, the presence of both a major reinforcing loop and three
counteracting loop processes, and policy questions about the effects of delays. In some cases
(e.g., Assignment 2.1 requiring an interpretation of the three stories of the basic model), the user
is required to post an answer to a personal electronic bulletin board. In others, users must
compare their own solutions with the suggested solutions, and post their assessment of
differences.
is
Suggested Solutions. Each set of simulation questions and assignments has a
corresponding set of suggested solutions. The user is encouraged to consult the suggested
solutions for each activity before proceeding to the next one. The culminating assignment
requires the user to post online an evaluation of their own solutions in light of the suggested
solutions. Figure 10 shows the Suggested Solutions page for Unit 2.
Suggested solutions are available below. However, before viewing them, keep in mind that this is
a self-guided learning activity. You are responsible to yourself.
It is highly recommended that you make a serious effort to answer each question and do each
activity fully. After you have made such an effort and as you complete each activity, then it can be
instructive to compare your work with the suggested solution. However, a premature "peek" may
undermine your learning progress by giving you a false sense of comprehension.
Here's how to proceed to gain maximum benefit:
(1) Review the original assignment (e.g., Sim A), repeat the simulation, and review the notes
and/or answers you wrote.
(2) Critique your original response. Does it still make sense? Modify it if necessary.
(3) Click the button below that corresponds to that same assignment (e.g., Sim A), and compare
your work with the suggested solution.
(4) Make notes to underscore the parts that you understood well and also to correct any
misunderstandings. In addition, note any comments that provide useful new insights.
Feel free to email me to discuss any part of this learning unit. (dwheat@wheatresources.com)
Suggested Solutions
Figure 10. Suggested Solutions Page for Unit 2
The Sim buttons in Figure 10 open text boxes containing suggested solutions for those
activities. The Assignment button takes the user to another page, showing causal loop diagrams.
Figure 11 provides an example of a solution page for a specific simulation activity, Sim A.
SSA1. Visible stocks: business inventories, business bank accounts, and other bank accounts.
Visible flows: production, sales, income, and spending. (Other stocks and flows "not visible" in the picture can be seen by
reading the story, and peeking inside the decision diamonds.)
SSA2. In equilibrium, the level of each stock will be unchanging, because the outflow from the stock will be exactly equal
to the inflow. The rates of flow can be very rapid, but equilibrium will be maintained as long as the inflows and outflows
are equal. Thus, equilibrium does not mean "nothing is happening.”
SSA3. The initial equilibrium value for each flow is $6 trillion/year.
SSA4. GDP (Gross Domestic Product) is a nation's annual production rate of all final goods & services, measured in
"dollars' worth of final goods & services per year." The "Production" flow in the model represents GDP. Production of
“apples and oranges" is difficult to measure. Over a year, the average Sales rate or the average Income would be
approximately equal to the Production rate. In practice, the value of the final Sales rate is used for official estimates of
GDP, and Income data provide an alternative, independent estimate.
SSAS. In disequilibrium, even a perfect measurement of Sales would not produce an accurate estimate of Production,
since the two flows would not be equal. Their difference could be calculated by measuring the change in the Business
Inventories stock over the year. The annual Production rate (i.e., GDP) would be equal to the annual Sales rate plus the
net change in Inventories during the year. Click "run" on the graph below to see this result.
1: production
6.004) 44
GDP is the production rate. In disequilibrium, the sales
rate is not equal to the production rate. However, the
difference can be found in the change in inventories.
‘
Years
That is how government statisticians estimate GDP.
Production = Sales plus Inventory Changes
Figure 11. Suggested Solutions for Sim A
Thus, the various simulation activities and the associated questions and assignments
involve users in the various stages of model-facilitated learning. Sim A illustrates problem
orientation. Sims B and C culminate in the causal loop diagramming exercise involving inquiry,
exploration, hypothesis formulation, and experimentation. Policy assessment is emphasized in
Sim D, where users consider the time delay circumstances that would be conducive to the
success of a particular fiscal policy action.
A central MFL concept is the principle of graduated complexity, and it guides the content
delivery process in two ways. As explained previously, MacroLab engages users with simple
subsets of the complete model early in the course, and progressively adds complexity. In
addition, within each week’s learning unit, the principle guides the developmental sequence of
assignments for that unit.
5. Research Agenda
System Dynamics vs. Comparative Statics. Section two suggested that the graphical,
comparative statics method of teaching macroeconomics fails in two respects. First, it does not
provide an effective learning pattern structure that facilitates student learning. Consider, for
example, the solution to this typical exercise, adapted from Radzicki (1993):
Find the new equilibrium natiional spending and income rates after an exogenous
investment shock of $10 billion, given an initial equilibrium at $400 billion and
an aggregate spending function defined by AE= .80*Y + I.
Figure 12 displays the graphical solution, derived by shifting the spending function up by
$10 billion. When combined with the algebraic determinatiion of the multiplier of 5 that is
implicit in the consumption coefficient of .80, the solution is a $50 billion increase in aggregate
spending. The difficulty is not in the analytic geometry or the algebra; both are straightforward.
The problem is the method’s failure to evoke a vivid mental image of what actually has to
happen in the economy—and what assumptions are necessary—before such a result could
prevail. That failure makes it difficult for students’ mental models to be renovated in any way
that is both useful and sustainable.
Aggregate
Expenditure
(AE)
AE = .80*Y + 90
Figure 12. Graphical Comparative
Exogenous Statics Solution to Investment Shock
Investment
increase
400 450
Aggregate Income (Y)
Section two also suggested that the comparative statics method fails to provide a realistic
picture of the transition from one equilibrium to another—a transition of paramount importance
to both ordinary citizens and policy makers. Consider, for example, the solution to the previous
exercise when graphed over time in Figure 13.
$500 9 Comparative
Ey
Initial !
Income J
cared SD: demand
side only
SD: combination of
demand & supply sides
25C 9
Initial tad
Investment
= 80 WS . Step
increase
0 T ¥ T 7
QO 30 60 90 120
Month
Figure 13. System Dynamics Solutions vs. Comparative Statics Solution
The dashed line represents the implicit comparative statics solution. Granted, most
economists and textbook authors acknowledge the issue of time, and they would not deliberately
suggest that the new equilibrium would be achieved immediately. Few, however, discuss the
practical significance of such delays, particularly those that strengthen counteracting feedback
loops.
Two system dynamics solutions to the same problem are also shown in Figure 13. The
first reflects only a “demand-side” perspective on the economy (i.e., the top half of the basic
model in Figure 3.) The dotted curve displays the effect of gradual adaptation of spending
patterns due to changes in income set in motion by the exogenous investment shock. The second
SD solution makes use of both the demand and supply sectors of the model. In addition to
gradual spending adaptations on the demand side, there are also supply-side adjustments to
changes in inventory stocks and desired labor stocks, and the solid curve displays the oscillatory
behavior resulting from the interaction of those counteracting feedback effects.
All three solutions are the same in the long run. However, the short-run business cycle
character of the second SD explanation is much more in line with actual experience. As a result,
it is more likely to renovate students’ mental models in an accurate, sustainable way. At least,
that is the hypothesis. Future research will test this hypothesis in formal experiments with
students receiving alternative methods of macroeconomics instruction, including the system
dynamics approach that is embodied in MacroLab.
Complete Framwork vs. “Biased” Framework. A concern over the efficacy of
macroeconomics instruction is not limited to strictly cognitive issues. One manifestation of
learning about economics is the attitude that one develops regarding appropriate government
economic policy and, in general, the role of government in the economy, a persistent conflict
with the economics community.’ Since Keynes legitimized the policy of “demand
management,” that conflict has sometimes been framed as “supply-siders” vs. ““demand-siders.”
Does "supply create its own demand" (Keynes’ coinage of Say’s Law), or is it the other way
around?
On the issue of government’s role, Say’s position is clear:
The encouragement of mere consumption is no benefit to commerce; for
the difficulty lies in supplying the means, not in stimulating the desire of
consumption; and we have seen that production alone, furnishes those means.
Thus, it is the aim of good government to stimulate production, of bad
government to encourage consumption. -
Such a policy perspective is in sharp contrast to Keynes’ conclusion:
I expect to see the State...taking an ever greater responsibility
for...investment, since it seems likely that the fluctuations [in private investment]
will be too great to be offset by any practical changes in the interest rate. ia
This issue raises a number of questions that might be answered, at least in part, by formal
experiments. In a recent pilot experiment, two small groups of graduate students at the
University of Bergen received “biased” macroeconomics lectures. The lecture for one group
focused on the demand side of the MacroLab model. The subjects studied a model that showed
economic behavior to be a function of decisions by spending agents in the economy, including an
example of successful demand management by an activist fiscal policy stance by government.
The implicit message was that demand creates its own supply, although that phrase was never
actually mentioned.”
The other group received instruction from a supply-side perspective, and the demand
side was taken as “a given.” There were no spending agents making demand decisions. The
sales rate was equated to the production rate. It was as if supply created its own demand,
'S The spread and adoption of an academic-based conventional wisdom would seem to be heavily influenced by the
paradigm of the profession and the means by which the mental models of both instructors and students are shaped by
that paradigm. Much of that influential shaping, of course, is done by textbooks. Paul Samuelson is quoted as
saying, “I don’t care who writes a nation’s laws...if I can write its economics textbooks.” (Skousen, 1997)
'° Say, chapter 15.
> Keynes, p. 164.
*! ‘The top half of the basic model was expanded in stages to include consumption, saving, investment, and
government taxation and spending. The supply sector was always visible, but there was no reference to that side of
the model. It served merely as a conduit for sales to be transformed into income via the production process. One
simulation run demonstrated how a sudden decline in consumption spending would be followed by declines in sales,
production, and income. Finally, government was added to the model and the effects of changing taxes and spending
were observed. In particular, due to a policy equation in the government sector, an increase in government spending
was triggered by the stagnating income, with the result that aggregate spending bottomed out more quickly and a
higher equilibrium value was attained. Potential implications of deficit spending (e.g., impact on inflation or interest
rates, crowding out of private investment, future tax requirements) were not discussed.
although that phrase was never mentioned.”
Before and after both lectures, the subjects provided answers to questions designed to tap
their attitudes towards fiscal policy as a stabilization tool, and their preference for government
spending or business tax cuts as a fiscal stimulus during a recession.
The hypothesis:
Subjects receiving “demand-side” instruction would exhibit more of a
Keynesian perspective relative to government’s role in the economy.
Operationally, in terms of the pre- and post-test questions, subjects in the “demand-side”
group would be more likely to (1) prefer “higher government spending” over “investment tax
incentives” as an anti-recessionary policy, and (2) exhibit a more favorable attitude towards the
use of fiscal policy as a stabilization tool. The results, while not statistically significant, were
consistent with the hypothesis.”
The pilot experiment was informal, highly exploratory, and used crude measurement
techniques. Efforts are underway to improve the experimental design and repeat similar
experiments with a variety of subjects. The primary hypothesis is that instruction presented in a
“scientific” manner will cause students to believe it when they see it, even when it is only half
the story. The next step is to test how such a transformed belief manifests itself. Do card-
carrying members of the supply-side group “see” the same data with a different conceptual lens,
and reach different conclusions about the state of the economy, compared to their demand-side
siblings? Do the two groups differ in the kinds of information they search for when seeking to
assess the state of the economy? Do the two groups evaluate a policy debate with a bias that
reflects their renovated mental models?
Most importantly, what benefits—in addition to a more accurate understanding of
economic systems—accrue to a third group of students who receive a balanced economic
education that seamlessly combines the demand side and the supply side?
7. Conclusion
MacroLab applies system dynamics and model-facilitated learning principles to the
problem of macroeconomics instruction methods. It is a series of interactive learning units
currently used as the organizing instructional tool in a distance learning course, and holds
promise as a supplementary tool in traditional undergraduate classroom settings.
Its effectiveness will be evaluated relative to comparative statics methods via formal
experiments. In addition, its capacity to provide a seamless presentation of both the supply- and
demand-sides of the economy will be tested.
Workshop participants will gain economic understanding by test-driving MacroLab, and
will gain ILE understanding by examining the content delivery process and the design process.
~ The supply-side of the model was expanded with detailed assumptions about labor supplies, average workweek,
and productivity. The concept of capital was presented as an embodiment of the productivity factor. Labor force,
workweek, and productivity were permitted to vary according to historical rates, and the subjects compared the
simulated results between 1900 and 2000 with the actual GDP and employment trends over that period.
* Details of the experiment are available on request.
References
Alessi, S. (2000). Designing educational support in system-dynamics-based interactive learning
environments. Simulation & Gaming, 31(2), 178-196.
Boucher, A. (1995). “Systems Modelling in Economics: Use of Object-Oriented Software.”
Computers in Higher Education Economics Review, 9:2 (virtual edition)
http://www.economics.|tsn.ac.uk/cheer/ch9_2/ch9_2p03.htm).
Becker, W. and M. Watts (1998). Teaching Economics to Undergraduates: An Alternative to
Chalk and Talk. Northampton, MA: Edward Elgar.
Bruner, J.S. (1963). The Process of Education, New York, NY: Vintage Books.
Cohn, E., S. Cohn, D. Balch, and J. Bradley (2001). “Do Graphs Promote Learning in Principles
of Economics?” Journal of Economic Education (Fall), 299-310.
Forrester, J. (1961). Industrial Dynamics. Waltham, MA: Pegasus Communications.
Forrester, J. (1968). Principles of Systems. Waltham, MA: Pegasus Communications.
Forrester, J., N. Maas, and C. Ryan (1980). “The System Dynamics National Model:
Understanding Socio-Economic Behavior and Policy Alternatives.” Technology Forecasting
and Social Change 9: 51-68.
Forrester, J. (1994). Policies, decisions, and information sources for modeling. In Morecroft, J.
and J.D. Sterman (Ed.) Modeling for Learning Organizations, Ed., Portland, OR:
Productivity Press.
Forrester, N. (1973). The Life Cycle of Economic Development. Cambridge: Wright Allen
Press, Inc.
Forrester, N. (1982). A Dynamic Synthesis of Basic Macroeconomic Theory: Implications for
Stabilization Policy Analysis. Unpublished PhD dissertation, Massachusetts Institute of
Technology. Cambridge, MA.
Godley, W. (1983). Macroeconomics. New York: Oxford University Press.
Kennedy, P. (2000). Macroeconomic Essentials: Understanding Economics in the News.
Cambridge: The MIT Press.
Maas, N. (1975). Economic Cycles: An Analysis of Underlying Causes. Cambridge, MA:
Wright-Allen Press, Inc.
Maas, N. (1980). “Stock and Flow Variables and the Dynamics of Supply and Demand” in
Randers (1980).
Milrad M., J.M. Spector, and P.I. Davidsen (2003). Model facilitated learning. In S. Naidu (Ed.),
Learning & Teaching with Technology. London: Kogan Page Limited.
Moxnes, E. (2000). “Not only the tragedy of the commons: misperceptions of feedback and
policies for sustainable development.” System Dynamics Review 16:4
Radzicki, M. (1988). “Institutional Dynamics: An Extension of the Institutionalist Approach to
Socio-economic Analysis.” Journal of Economic Issues, 22:3, 633-665.
Radzicki, M. (1993). “A System Dynamics Approach to Macroeconomics.” Unpublished paper
delivered at the Univeristy of Bergen. March, 1993.
Richmond, B. (2000). An Introduction to Systems Thinking. Hanover, NH: High Performance
Systems.
Skousen, M. (1997 The Making of Modern Economics: The Lives and Ideas of the Great
Thinkers. New York: M.E. Sharpe.
Spector, J. M., & Davidsen, P. I. (2000). Designing technology enhanced learning environments.
In B. Abbey (Ed.), Jnstructional and cognitive impacts of web-based education. Hershey, PA:
Idea Group.
Sterman, J. (2000). Business Dynamics: Systems Thinking and Modeling for a Complex World.
Boston, MA: McGraw-Hill Companies.
Stigler, G. (1963). “Elementary economic education.” American Economic Review. 53:2.
653-59.
Walstad, W. B., and S. Allgood. (1999). “What do college seniors know about economics?”
American Economic Review. 89:2. 350-354.
Wood, W.C. and J.M. Doyle (2002). “Economic Literacy Among Corporate Employees.”
Journal of Economic Education (Summer), 195-205.
