Simulation-Based Learning Environments to Teach Complexity:
The Missing Link in Teaching Sustainable Public Management
Michael Deegan
US Amny Comps of Engineers
Krys Stave
University of Nevada Las Vegas
and
Rod MacDonald
Initiative for System Dynamics in the Public Sector
David Andersen and Minyoung Ku
Rockefeller College of Public Affairs and Policy
Eliot Rich
School of Business
University at Albany--State University of New Y ork
Abstract
While public-sector management problems are steeped in positivistic and socially
constructed complexity, public management education in the management of complexity lags
behind that of business schools, particularly in the application of simulation and simulation-
based leaming. This paper describes our development of a Simulation Based Leaming
Environment that includes a coupled case study and SD simulation surrounding flood protection,
a domain where stewardship decisions regarding public infrastructure and investment have direct
and indirect effects on businesses and the public. The Pointe Claire case and Coastal ProtectSIM
simulation provide a platform for policy experimentation under conditions of exogenous
uncertainty (weather and climate change) as well as endogenous effects generated by structure.
We discuss the model in some detail, and present teaching materials developed to date to support
the use of our work in public administration curricula. While leaming and outcome evaluations
are not complete, we believe that he effectiveness of this approach will be demonstrated.
Overview
There is a new challenge facing public management education—to teach public managers
to handle a broad range of novel situations characterized by complexity when dealing with an
emerging class of problems that we dub “sustainable” public management problems. This paper
first gives a quick overview of the current state of public management education, poses a
preliminary multi-dimensional concept of complexity that encompasses both positivist and social
constructionist view of complexity, and proposes a broad design for simulation-based leaming
environments (SBLEs) to teach in this complex domain. We next present an example of one such
SBLE—the Pointe Claire Coastal Protection Case, a case focusing on the decisions of a Regional
Coastal Planning Commission on the Mississippi Coast faced with the dual threat of current
storm damage from hurricanes such as those already hitting the coast (e.g., Katrina) as well as
31® Intemational System Dynamics Conference, Cambridge, MA (2013) a
the future probable threats of enhanced damage due to global warming. Finally, the paper
discusses how this SBLE was implemented in a first class on modeling methods in the
Rockefeller College's core MPA program and presents some preliminary results from instructor
attempts to evaluate the instructional technology as well as student leaming in this complex
domain. The paper concludes with reflections on future research needed in this area.
Part I: Traditional Public Management Education and Complexity
This section reviews the current state of public management education and briefly
discusses complexity in decision making, suggesting a simple but comprehensive taxonomy of
complexity in public policy decisions.
The Current State of Public Management Education. The current public management
education has relied heavily on the traditional classroom leaming which assumes that knowledge
and skills which are needed for sustainable public management can be transferred from the
instructor to the students through readings and lectures (Comfort & Wukich, 2013). According to
Comfort and Wukich (2013), it is true that even the majority of courses on crisis management
currently offered in MPA programs are designed and managed based on this traditional principle
of teaching and leaming environments. However, the rapid change in the public policy decision
making environments, especially, the increase of complexity has brought the need for exploring
a new set of qualities, which are expected to public managers, and the ways to nurture these
qualities in MPA programs.
As the core qualities of successful public managers, the National Association of Schools
of Public Affairs and Administration (NASPAA) have suggested MPA programs to pursue the
five competencies: the ability (1) to lead and manage in public govemance, (2) to participate in
and contribute to the policy process, (3) to analyze, synthesize, think critically, solve problems
and make decisions, (4) to articulate and apply a public service perspective, and (5) to
communicate and interact productively with a diverse and changing workforce and citizenry.
However, the detailed components of each type of competencies are not defined by the
NASPAA. Rather, the NASPAA encourages institutions that rm MPA programs to define the
meaning and sub-components of the competencies—i.e., knowledge, skills, and abilities.
Following this idea, as an effort to improve the competitiveness of the MPA program, the
Rockefeller College of Public Affairs and Policy (the Rockefeller College, for short) has
elaborated the NASPAA’s five core competencies by group brainstorming among MPA faculty
and has applied the sophisticated understanding of the competencies to the current MPA core
courses of the Rockefeller College (See Appendix A).
A Proposal for Thinking about Complexity. Public managers and policy makers in the
% century are required to manage complex systems whose boundaries spill over agency,
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 2
jurisdictional, and sector boundaries, dealing with a great deal of uncertainty. Ever since
Lindblom (1959) first brought up complexity as a new topic, the literature has reviewed many
features of such “wicked problems” framing issues about how to deal with complexity in the
public sector. Although various approaches to conceptualizing complexity do exist, much less
attention has been paid to methods and approaches for teaching and leaming in and about
complex systems in public management settings.
Here, we suggest a taxonomy of “complexity in public policy decisions” encompassing
positivistic and interpretive features of systems complexity. This taxonomy, shown in Figure 1,
classifies the features of complexity in public management settings largely into two dimensions:
(1) positivistic complexity, which is a bundle of objectively observable and measureable features
that make public policy problems difficult to manage (such as decision-making in the face of
stochastic uncertainty or feedback complexity within complex systems models); and (2)
interpretive conplexity, which results from the diverse interactions of multiple stakeholders with
often competing points of view, leading to intra-group, organizational, or political conflicts.
Figure 1. A Taxonomy of Complexity in Public Policy Decision
Technical goal issues Stochastic Uncertainty
(Use optimization) {Use Decision Trees)
Thinking Through
ime
Positivist Liposrnlc Glare
Complexity (Use dif ference equation:
(Model as ‘System Dynamics) Stock-and-Flow
Micro-World) Thinking
Feedback
Detail Complexity
Complexity (Use MAU odes) Complexity
In Public
Policy ‘Other. Individual Value
Decisions Differences
Electoral Politics er
Stakeholder Complex
Organizational
Socially Constructed ‘Values and Norms
Complexity (Model as
Boundary Object) Complex, Diverse Equity vs. Efficiency
‘And Competing
Folicy Goats Free Market vs.
Interventionist
piieenbrta Orientation
Other Institutional constraints
Ete.
Part II: The Pointe Claire Coastal Protection Planning Exercise— Toward a Simulation-
Based Learning Environment.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 3
Given the taxonomy, shown in Figure 1, we designed and built a SBLE to teach how to
manage multiple dimensions of complexity based on the “double-looping leaming model” that
Stemman (1994) suggested as a teaching and leaming model of complexity. This curriculum for
teaching complexity within public policy decisions makes use of a simulation-based large-scale
case focused on “Disaster Preparedness on the U.S. Gulf Coast in the face of Global Wanning.”
The complete curriculum consists of a realistic system dynamics simulation model of the impact
of hurricane grade stomms on a typical coastal community plus a series of exercises that focus on
stakeholder complexity and decision making within a community-based goveming board tasked
with planning for such storms in the face of future-possible global warming threats.
The Pointe Claire Regional Coastal Planning Exercise wes a multi-component
simulation-based exercise that spanned over ten weeks of activity in a core MPA class in
modeling. The purpose of the exercise was two-fold—(1) In substantive tems, to teach students
to use a complex simulation model as a tool to understand a multi-faceted set of interactions and
come up with robust policy conclusions, and (2) In terms of the policy process, to teach students
how to use complex models to help groups of public policy stakeholders come to
around policy goals. The class exercises were built around a system dynamics simulation of
coastal protection dynamics, CoastalProtectSIM.
e Students engaged in an in-class exercise working with the C-ROADS simulation, a high
fidelity simulation system used to forecast impacts of CO2 emissions on global warming
over a 50 year plus time horizon (See Appendix B-1).
e Students drafted a memo detailing a way to use the C-ROADS simulator as part of the
coastal protection planning process in the Pointe Claire Region (see Appendix B-2).
e Students participated in a group model-building exercise in which the class mapped out a
system structure similar to the structure of the CoastalProtectSIM (See Appendix B-3).
e Students participated in two computer lab exercises where they formulated portions of the
CoastalProtectSIM model to become more familiar with how the model was formulated in
detail (See Appendix B-4).
e Students participated in role playing exercises in classroom discussions so that they gained a
better feel for how key stakeholders took positions on coastal protection.
e Working in small groups, students “solved” the policy problem and drafted a policy memo
with a supporting set of PowerPoint slides indicating what they found to be the “best” policy
solution and why (See Appendix B-5).
e Students did background reading in three related perspectives on public policy formation—
(A) readings on stakeholder analysis and management in the policy process, (B) readings in
the creation of mini-publics as a way to achieve policy consensus, and (C) readings on
organizational leaming and systems thinking as goals of networks or organizations working
in the public policy field.
e Students drafted individual papers using the three sets pf background readings in public
policy plus their work with the simulator (See Appendix B-5)
Part III: Details of the C oastalProtectSIM model
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 4
Coastal Protect Sim was developed to replicate several types of real world complexity:
(1) time delays in constructing costal protection; (2) cost sharing challenges for construction and
annual maintenance; (3) impacts of costal land development on natural bariers; and (4) the
timing of benefits and costs in net present value calculations for long range coastal flood risk
Planning. The model uses a random seed to create micro-worlds, whereby the probability of any
particular storm may generate a surge large enough to exceed natural and man-made protection.
In addition, a global warming scenario is built into the model that allows for the amplification of
the storm surges based on severity of storms and sea level rise. Costs associated with mitigation
and benefits from damages avoided are calculated in terms of their net present value at the OMB-
required 7% discount rate. Coastal Protect Sim requires the decision maker to determine whether
the long term benefits are worth the investment of short and intermediate term mitigation
measures. The temporal boundary for the model is 40 years to allow for long term and short
term tradeoffs to be explored. In this section we begin with a description of the model structure.
We then provide base run behavior for three mircowords and two climate change scenarios. The
section concludes with a description of several policy runs anda discussion of tradeoffs for each
strategy.
Coastal Protect Sim Model Description. Coastal Protect Sim (Figure 2) has three model
sectors and two model structures for accounting benefits and costs. All five areas of the model
are discussed in this section of the paper: (1) structural mitigation protection; (2) land
development and natural barriers; (3) storm intensity and climate change; (4) costs associated
with damages and mitigation measures; and (5) benefits from cumulative tax revenue. Table 1
provides a legend for the causal map to help the reader identify each of the five variable types
discussed in this section of the paper.
Table 1; Legend for Coastal Map causal map
Causal link | Coastal Protect Sim Model Structure
color
Blue Policies to mitigate d and minimize recovery costs
Brown Natural barriers to protect the community
Purple Storms and climate change
Red Di: d and mitigation costs
Green Benefits from tax revenue and d avoided
Coastal Structural Mitigation
Starting in the upper left comer of the model, Coastal Protect Sim captures the
connection between the planning and implementation of structural coastal barriers. Community
decision makers identify the desired level height of protection and project start time. However,
the time to complete the plan formulation process is not within the control of the local decision
maker. As the Corps of Engineers currently goes through a “Transformation” period, it is moving
towards an accelerated planning process to address concems the process is currently too
expensive and too lengthy. The accumulation Built Protection in Planning reflects the delay
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 5
between desired levels of mitigation and the time it takes to complete the reconnaissance and
feasibility studies. In the Corps budgeting process, completed plans lead to Built Protection
Being Sited through Preconstruction Engineering and Design (PED) investigations, which is an
intermediate step before formal construction. The final accumulation Finished Build Protection
is based on the rate of construction for protective structures along the coast. In the base run of the
model, the total delay for these three stocks is 10 years, which corresponds to the average delay
time in the USACE planning and construction process.
Projects that have been completed increase the Total Coastal Protection which reduce the
amount of storm surge the community experiences directly (Inches Above Protection Margin of
Safety). The model assumes a threshold where storm surge will produce some degree of property
damage. As storm surge rises above the total protection on the coast, the Effect of Storm Surge
on Damage increases to a potential Maximum Damage Per Acre Per Storm, which has been set
for the base run at maximum of $100K/acre. Current Storm Damage is also influenced by the
building codes effect on damage, which represents a policy whereby floodplain managers are
able to successfully implement codes to guarantee lower levels of property damage during the
next storm event.
If the Current Storm Danuge is higher than the protection provided by structural policies
or strict building code enforcement, the resulting percent damages indicate the extent of damages
in the community. If this percent is relatively large, the landowner willingness for buyout will
increase as well. It is conceivable landowners would be willing to relocate during the recovery
period, thus creating open space and increasing the level of Undeveloped Coastal Land and
reducing Developed Coastal Land. The potential balancing feedback loop suggests an
opportunity to minimize future damages. Altematively, a zoning regulation can be enforced to
restrict development, which would help to guarantee the balancing loop maintains its goal
seeking behavior. There is a caveat with respect to the link between percent damaged and
landowner willingness for buyout. Coastal Protect Sim has model structure (hidden in this view)
that activates federal disaster assistance in very large disasters, which may reduce the willingness
to relocate in certain cases.
Natural Barriers
The level of Undeveloped Coastal Land (center of Figure 1) acts as a natural barrier to
protect against storm events. As this level increases, its impact on Natural Protection increases,
which enhances the natural environment during major storm events. Communities that maintain
large sand dunes between developed property and the ocean, as well as sustainable beaches
solutions to import or relocate sand on the shore have more protection during hurricanes and
major storm events. The natural barriers combine with structural protection to increase the Total
Coastal Protection, which as previously discussed minimizes storm surge and flood damages.
However, this added protection also increases the perceived protection in the community. A high
perceived safety for development acds pressure on the community to expand and develop on the
shore. As the impact of safety on development increases, it may add to development in the
community. This balancing loop could play a dangerous role in the model, especially in
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 6
circumstances with a long delay in between actual and perceived safety in the community.
Stonss and dinate change
The Coastal Protect Sim model operationalizes storms through two concepts: mean storm
surge and volatility. Storm Volatility is formulated as a Random Normal with a range of -50 to
400 inches, Seed, which uses an initial storm volatility of 24 inches that can have an impact of
global warning to amplify the volatility. The Random Seed effectively selects one possible
future microwords. During model testing several seeds were selected to represent the more
interesting and challenging future worlds. To account for climate change, the model associates
an impact on volatility by Temperature Rise by 2052, with an associated percent increase in
volatility per degree rise. In the base run the temperature rise is set at zero. The mean max storm
surge is set at 108 inches in the base run with the potential to increase based on the impact of
global warming on mean max surge. In the base run, the percent increase associated with each
degree in temperature is 5%. Sea Level is a third contributing factor to storm surge. It is set at
zero in the base run. The fourth and final contributing factor is the Effect of Storm track on surge,
whose purpose is to add a layer of uncertainty in the model. That is, not every storm is perfectly
predicted. Total Storm Surge is the result of Mean Max Storm Surge, Storm Volatility, Sea Level
Rise, and the effect of storm track In most cases, Inches Above Protection is negative, which
results in a zero effect of storm surge on damage. However, in those cases where this value is
above zero, the effect can be rather large. For example, the initial coastal protection is slightly
above 150 inches, so any run where the seed produces a value greater than 150 inches will result
in potential damage. In the base run of random seed 20, inches above protection margin of safety
is positive 3 times in the 40 year run.
There are two types of costs recorded in Coastal Protect Sim. First, the model records
costs associated with the implementation of mitigation policies. For example, as shown in the
upper left comer of Figure 1, the model records current planning costs, current siting costs,
current construction costs, and maintenance costs at an annual rate which feed into a Net Present
Value of Current Adjusted Costs. There are major financial challenges for many communities
who wish to participate in structural mitigation measures on the coast. Even after project
construction has been completed, communities must participate in cost-sharing for the
maintenance of these projects. In the model, the costs are recorded and discounted at the OMB
required rate of 7%.
The second cost in the model is from property that has been purchased or reclaimed by
the state. Once again, even in cases where the federal govemment supports a buyout of local
property, there is usually some level of cost sharing on the part of the non-federal partner. In
addition, there are costs associated with the implementation of strict building code policies,
which carry a direct burden to the homeowner. Finally, the cost to recover a community after
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 7
disaster is recorded as stock of Cunulative Storm Damages. Taken together, these costs
determine the level of successful (or failure) fora given set of mitigation policies.
Benefits fromtax revenue and damages avoided
Benefits are shown in the lower left comer of Figure 1. Coastal Protect Sim allows the
decision maker to implement a tax policy to offset the community cost-sharing burden. To be
clear, taxes collected are for a single purpose. Taxes to be collected for other issues, such as
crime, education, and infrastructure are beyond the scope of the model. The model calculates a
desired tax rate based on the aforementioned Net Present Value of Current Adjusted Costs. The
total land value is used to then determine an appropriate tax rate. With that said, the user must
careful not to overburden their taxpayer, as unreasonable taxes could have an adverse impact on
sustainable development.
The variable Cumulative Damages Avoided is calculated based on a model structure that
replicates the one presented in Figure 1, with one important distinction. Essentially, there is a
second model which runs without any govemment involvement. The resulting damages from the
“no govemment’ model is compared to the policy mms in the “govemment’ model. The
difference between Cunulative Storm Damages in these models is recorded as Cumulative
Damages Avoided. Cumulative Damages Avoided is added to the revenue generated from taxes
for a total Cumulative Benefits and Damages Avoided. The difference between this total and the
Cunulative Costs and Damages is recorded as Total Net Benefits.
Model Behavior
The model generates storms and storm surges over the course of a 40 year period. The
storms are randomly generated and a percentage of the storms may exceed the man-made and.
natural barriers and cause storm damage. Users read a case history about the leadership
challenges facing the county executive of Point Claire. From their understanding of these
challenges, users develop a flood risk management strategy that technically, financially, and
politically feasible. The model is used to test strategies under various scenarios and communicate
the results.
The following selection of model runs highlights different types of uncertainty and tradeoffs
unique to this particular policy domain.
The Base Runs
The base run for each random world has the same set of assumptions. Pointe Claire
begins as a community with minimal flood risk management policies in place. It relies heavily on
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 8
natural barriers to provide flood protection. Therefore, the base nm for each random seed
highlights different types of storm “challenges”. A policy mix that performs well under one
random seed may not achieve the same level of success under another random seed.
Total Net Benefit
‘ih!
Pe eee
-4B
2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052
Tine (Year)
Total Net Benefit : 48 —+——+——_4+—_ Total Net Benefit: 20 -—=-=—=-=3=
Total Net Benefit :10 ————2——>-
Random Seed 48; The base run in random world 48 experiences four events beyond the
protection of its natural barriers. The first event occurs midway in the run, with a second event
10 years later. The final two events are rather small and occur at the end of the base run.
Random Seed 10: In the base run of random world 10, the commumity is hit with three events in
arow. However, all of these events occur rather late in the run, starting at approximately year 30.
Random Seed 20: In the base run of random world 20, the community is hit with an event almost
immediately. The next event beyond its natural barriers occurs approximately 25 years later. A
third event occurs another 10 years later, with each subsequent event slightly less damaging than
the previous.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 9
Global Warming
The model was run several times to reflect different climate change scenarios. Three
examples under random seed 20 are presented in this paper. The base run with sea level rise at 6
inches has some impacts in the later years of the run. The total cumulative damages are similar to
the base rm. A second global warming run with parameter change for temperature rise of 3
degrees (5% surge per degree) results in relatively higher damages toward the end of the rm. A
final global warming run in random seed 20 had a 3 degree temperature rise with a 10% surge
per degree. This global wamming test results in a change in both frequency and severity of
damage, with several more events creating damage in the later years. This final test shows
cumulative damages nearly double the size of the base run.
Total Net Benefit
f -3B 7 =
, Sy
2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052
Tine (Year)
Total Net Benefit : 20. ————4t————_4 Total Net Beneit ; 20temp3a5Spr ——
Tota Ne Bendit : 2016 ——2 $$$ Total Net Benefit : 20temp3dt] Oper
Policy Runs
The Coastal Protect Sim model has several types of policy altematives to explore. A
description of each policy, with recommended policy values along with default values in the base
tun is described in Table 2. The recommended values are merely suggestions to decision maker
to provide some boundaries and make it easier to keep track of many policy mix combinations.
The contents in Table 2 were provided to the decision makers to make them aware of all policy
options in Coastal Protect Sim model.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 10
Table 2: Coastal Protect Sim Policies
Policy Description Default/
Parameter Recommended
Policy Values
Height of | Built protection for Pointe Claire results in projects such as | Default: 0
Protection seawalls, beach replenishment, and banier island
replenishment The height of built protection adds to the
commumity’s existing natural environment protection. It takes
approximately 5 years to complete the initial planning studies | Policy values:
and at least another 5 years to complete the construction | 0,18, 24, or 36
project.
The height of man-made protection will determine the
construction and annual maintenance costs. In the real world,
cost-sharing requirements make it difficult for some
communities to participate in agreements with the Corps.
Therefore, both construction and maintenance costs should
be considered to determine the appropriate height of
protection.
Tax Rate for | There are several costs to consider in the model: costs for | Default: 0
Protection Plamning, construction, and operations & maintenance. Taxes
can cover the non federal share of these costs. If you set the
tax rate higher than that cost of the project, your tax revenue
benefits will accumulate. Be careful. If you set the tax rate| Policy values:
too high, your taxpayers may revolt against you! between 0
and .002
Automated You may notice it is difficult to set the tax rate just right. | Default: 0
Tax Rate Instead of setting the tax rate for protection, you may opt to
use the automated taxes feature. When this feature is | Policy values:
activated, you will be guaranteed to collect taxes exactly at
the cost of your height of protection Oorl
31° Intemational System Dynamics Conference, Cambridge, MA (2013) 11
Building
Code Policy
One way to avoid damages without clearing homes from the
floodplain is to develop strict building codes for
floodproofing and elevating structures above the base flood
elevation level. Building codes won't eliminate all of the
damage during a storm. Set the building code policy to any
number between 0 and 1. This will be the percent of
structures (the goal) you hope to be in compliance with your
codes. Also, keep in mind that building codes come at a cost
to the property owner.
Building codes should be considered as part of a holistic
flood risk management strategy. Since costs will be
immediate and benefits will potentially occur only after
damages are avoided, the year in which the building policy is
implemented plays an important role in both cumulative costs
and damages.
Year of
Building
Code Policy
The enforcement of building code policies make structures
less prone to storm surge damage. These policies reduce
damages and save money when storm surges exceed the
height of protection. Building codes increase property
Maintenance costs on homeowners and businesses. Unlike
seawalls and large structural mitigation projects, building
codes place more financial responsibility on the individual.
Floodplain managers are accountable for the implementation
of these policies. These policies are rather important, as
FEMA Commumity Rating System (CRS) points and
National Flood Insurance Program (NFIP) discounts depend
on their successful implementation.
Default: 2020
Policy values:
between 2012
and 2052
Buyout or
relocation
policy
Buyouts, relocations, and reclamation policies remove homes
from the floodplain. Pointe Claire does not have the
resources to remove homes before a disaster strikes.
However, if you decide to implement a buyout policy,
landowners will be inclined to accept a buyout during major
events. They are less likely to accept a buyout during smaller
events. Federal programs such as the FEMA Hazard
Mitigation Grant Program help minimize reclamation costs
on the local community. The buyout policy in Coastal Sim
represents the percent of properties offered a buyout during
the next event.
Default: 0
Policy values:
Otol
31® Intemational System Dynamics Conference, Cambridge, MA (2013)
12
Year of
Buyout Policy
Select the year when the buyout policy goes into effect. It is
assumed that once the policy goes into effect, buyouts will be
offered for every event after that year Keep in mind, buyouts
will not be offered immediately. Buyout offers are only
extended to residents after events where Pointe Claire incurs
damages.
In this model, if not storm occurs after the buyout policy,
then no land is reclaimed.
Default: 2020
Policy values:
2012 to 2052
Zoning
Regulations
Each community faces a delicate balance between zoning for
“open space” and zoning for land development. Zoning
regulations prevent new development in flood prone areas.
Development in Pointe Claire can change over time based on
policy decisions. The value of the zoning policy is the
percent of development prevented. Keep in mind that strict
zoning policies lower the tax base in Pointe Claire. A lower
tax base lowers the amount of tax revenue that may be
collected to offset the cost of structural protection projects.
Therefore, zoning regulations could generate costs for
remaining homeowners.
Default: 0
Policy values:
Otol
Zoning Policy
year
Select the year when zoning policies may go into effect.
Zoning policies take effect immediately. Zoning regulations
should be considered as part of a holistic flood risk
Management . The year in which these policies go
into effect may not lead to immediate implementation.
Therefore, the year the zoning policy is implemented is
important policy and determined by the user
Default: 2020
Policy values:
2012 to 2052
A few policy runs have been selected to illustrate some of the policy options in the
model. The timing of costs and benefits present a formidable challenge to the decision maker, as
some policies only yield a strong net present value due to events in the later years of the model
run. Other policies could be hindered by factors beyond the commumnity’s control, such as delays
in the Corps planning process. Yet other policies show that no single approach is enough to
sustain development in this coastal community. The handful of policies selected for discussion in
this paper highlight challenges in policymaking and strategic communication, as each policy mix
holds a unique set of tradeoffs. For simplicity, each policy described in this section uses random
seed 48 in the base run.
31® Intemational System Dynamics Conference, Cambridge, MA (2013)
13
Structural Protection
Total Net Benefit
4B
Ps sglosy
| 0 & x ngt
A ae
-4B + + + + + +——_-4
2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052
Tine (Year)
Total Net Benefit :48 —+ - ‘ - - - - - ‘ - 4
Total Net Benefit : 48protect24
Total Net Benefit : 48protect24delay
The policy num for structural mitigation is interesting for two reasons. First, on the surface
the policy appears to be rather successful against the base case. Whereas the base run results in
final total costs to the commumity in excess of 3 billion dollars, the coastal protection from
engineered solutions yields a net benefit in damages avoided of nearly 2 billion dollars. Recall
random seed 48 has four events that exceed the community’ s natural barrier protection. After the
first event, the policy solution does not produce enough benefit to warrant the cost of the project.
However, as the model continues to mm, it is clear the benefits exceed the costs. Also important
to note, the Corps of Engineers uses a 50 year life for most of its planning studies. The second.
interesting observation on this policy is its sensitivity to delays in the system. The model was run
a third time to reflect an additional five year delay in the coastal protection project. This delay
results in rather severe damages in during the first event. In fact, total net benefits of the policy
just barely rise above zero, which is due to avoided damages in the last year of the rm. This
example shows two ways Coastal Protect Sim model can help decision makers identify and
discuss the uncertainty and timing of costs and benefits in flood prone commumities.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 14
Building Codes
Total Net Benefit
0
=a -2B 2 = > 2
A + + + + +
=y
-4B
2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052
Time (Year)
Total Net Benefit : 48 —+ 4 + + + 4 4 4 + + 4
Total Net Benefit : 48buildingcode
Total Net Benefit : 48buildingcode2012
While a “building code only” approach does not quite produce robust outcomes in
random woud 48, the policy highlights an interesting challenge for decision makers. For this
policy run, the community sets building codes at a goal of 100% compliance. To reflect the
political capital needed to get such level of compliance, the policy goes into effect in 2020.
Compared to the base run, the delay in implementation results in damages similar to the base
during the first major event. However, with each subsequent event most of the damages are
avoided. A third run of the model with an earlier implementation start date (2012) is a vast
improvement on the same policy with a slower rollout strategy. In this run, building codes are
fully implemented by the first major event and most of the damages are avoided. However, since
building codes have a burden on the individual property owner, the result is a net zero benefit to
the community.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 15
Buyouts and Zoning
Total Net Benefit
° pt
g
=a -2B aH
a Ot et tt
Ty
-4B
2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052
Tine (Year)
Total Net Benefit : 48 + + + f 4 + + ft + 4 |
Total Net Benefit : 48buyoutOnly
Total Net Benefit : 48buyoutzoning
Perhaps the most realistic feature of the Coastal Protect Sim model is the fact that no
single policy serves as the magic bullet in flood risk management. Flood risk management
requires a holistic systems view of the problem. This is certainly true at the Corps today, where a
new focus has been placed on coordinating structural and nonstructural measures. The “buyout
only” approach barely outperforms the base case in random world 48. There are two inherent
challenges with this policy. First, damages must be large enough for property owners to be
willing to accept a buyout, but not too large to receive federal assistance to recover status quo ex
ante. Second, buyout policies alone do not remove the pressure to redevelop on the coast. A third
tun of the model with buyouts and zoning policies prove to be a more sustainable solution. While
net benefits are not quite above zero by the end of the nm, these policies show that a holistic
approach has more potential benefit. That is, by placing pressure on both the inflow and outflow
of the land development sector stocks, the policy mix helps to contain future damages.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 16
Figure 2: Coastal Protect Sim Model Structure
Peroent Increase in
‘Suye Per Dogee Rise
/ Temp Rise by
_ 2052
“
Inpactof Global
‘Wamirgoon Mean Max
3 ‘Sue
31% Intemational System Dynamics Conference, Cambridge, MA (2013) 17
Part IV: A Survey-Based Preliminary Evaluation of the SBLE
The goal of the SBLE is to enhance the NASPAA’s five core competencies which are
necessary qualities to become competent public managers who can deal with complex public
policy problems by providing MPA students with a comprehensive and well-designed
complexity leaning environments. To measure the students’ self-assessments on the potential
effectiveness of the SBLE, the Pointe Claire Coastal Protect Case on the enhancement of the five
core competencies, we conducted a survey to 44 MPA students who took the two classes at the
Rockefeller College in the 2012 fall semester, in which we administered the SBLE case.
The survey questionnaire is designed to measure respondents’ self-evaluation on (1) the
effect of the SBLE case on the increase of students’ interest in leaming complexity and (2) the
effect of the SBLE case on the enhancement of students’ “self-assessed” five core competencies
by 7-point Likert-type scale, ranging from agreeing on the statement “not at all (1)” to a very
great extent (7)” (See APPENDIX D). The survey questionnaire consists of three sections: (1)
Section I including questions asking about students’ perceptional and emotional experience with
the SBLE case regarding how the leaming package affects leamers’ motivation to leam
complexity; (2) Section IT including questions asking about students’ perceptions of how much
the class activities on the SBLE case served to improve the five competencies; and (3) Section III
including questions asking about students’ demographic information and the past and current
education and work experiences.
Part V: Implications and Future Work
This experience demonstrates the value of using simulation-based leaming environments
to build a more complee and useful understanding of public policy complexity in public
management education. It explored the possible effectiveness of a curriculum designed to teach
complexity using a simulation-based large-scale case study coupled with group exercises
intended to emulate complex interactions between key stakeholders in the case. We believe that
such simulation-based leaming exercises can and will have implications beyond the MPA
classroom, providing leaming tools for public managers at many levels of govemment.
Simulation-Based Learning Environments. This case is an example of a simulation-
based leaming environment (SBLE). A simulation-based leaming environment is a package of
materials and scaffolded exercises designed around a simulation model. SBLEs can be designed.
for use with varying degrees of facilitation, ranging from stand-alone packages that require
almost no extemal guidance to exercises used in classroom settings with significant instructor
facilitation. Simulation-based cases are widely used in business education, but are relatively new
in public management education (see J PAE article on teaching with simulations).
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 18
The use of the Pointe Claire SBLE described here supports the potential of SBLEs to teach
complexity in public management education, The experience raised a number of questions and.
indicates directions for further research and SBLE development. The evidence from this case is
promising, but the approach needs to be implemented in more classes and in other related
disciplines. It needs to be formalized and subjected to careful evaluation and analysis from one
or more rigorous frames of analysis.
This case can be viewed as the pilot phase in a larger research agenda examining the value of
SBLEs for improving public management capacity for working with complexity. In the pilot
phase, we explored the broad questions: Can SBLEs deliver the complexity leaming outcomes
needed for building capacity? And: What is the added leaming value of an SBLE beyond
traditional teaching?
Analyzing the case raised questions about how to revise it to make it more effective, and, more
generally, what general insights could be applied to developing similar SBLEs for other leaming
audiences. For example, we see applications for teaching sustainability in many fields including
business, environmental studies, and disaster management, for example. More work needs to be
done to understand how to improve the approach, measure the leaming outcomes, apply it across
disciplines, and implement it with different types of leamers. Some of the research questions
include:
(1) What is the best way to evaluate participant leaming about complexity? (2) What is the
effect of the SBLE on participant leaming? (3) What features of the SBLE most effectively
promote leaming about complexity—both complexity in the physical system and complexity in
human small group decision making?
How can this kind of SBLE best be used across the range of potential leamers? How do leamer
characteristics affect leaming outcomes of SBLE use?
These types of SBLEs have the potential to secure thoughtful public engagement in sustainable planning
across a wide range of domains that share features in common with coastal protection. The case described
here focused on public management students, but the approach has potential for use with other groups,
including community stakeholders. It will improve the ability of the public management workforce to
engage the public in decision making about sustainable futures.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 19
Appendices for
Simulation-Based Learning Environments to Teach Complexity: The Missing
Link in Teaching Sustainable Public Management
The full appendices for this paper are contained in a supplemental file attached to the conference
proceedings for the 2013 System Dynamics Research Conference. Below is an abbreviated table
of contents indicating the broad contents of these appendicies:
Appendix A: Restatement and Elaboration of 5 NASPAA Core Competencies by MPA
Faculty of the Rockefeller College. ‘This appendix contains a statement of the 5 NASPAA Core
Competencies plus additional elaborations on these competencies as adopted by the faculty of
Public Administration and Policy at the Rockefeller College, University at Albany.
Appendix B.1 Global Warming and the Pointe Claire Regional C oastal Planning
Commission— Part 1.4. This appendix sets up the basic conditions of the assignment for the
rest of the class. The first part of the assignment is directed toward a class exercise where the
students interact with the C-Roads Climate Change Simulator
Appendix B.2 Global Warming and the Pointe Claire Regional C oastal Planning
Commission— Part 1.B (class exercise with C-Learn Model). Thus appendix contains the
handout that was used in class for the C- Roads Simulator exercise.
Appendix B-3: Roles for Global Warming and the Pointe Claire Disaster Preparedness
Case. The Pointe Claire Case had a number of class-based role playing exercises. This
document describes the basic roles that students assumed during the class exercises.
Appendix B-4: Notes for Formulating a Simple Difference Equations Model for Pointe
Claire C oastal Protection. The class had a homework assignment requiring them to formulate
some of the key dynamic structures within the CoastalProtectSIM model. This is a worksheet
that groups of students using during a class lab to get started on the homework assignment.
Appendix B-5: Global Warming and the Pointe Claire Regional C oastal Planning
Commission— Part 2. The final assignment had two main parts. Working in small groups, each
team crafted a short presentation that used the simulator to create a “solution” for the Pointe
Claire Regional Planning Commission. In addition, as an individual assignment, each student
drafted a policy memo addressed to the Director of the Commission giving her advice on how to
use a formal simulation model to support policy formation. This document sets up both of those
assignments as well as directs students to background readings on stakeholder analysis and
Management, material that had been previously assigned in another MPA core class.
Appendix C: End of Class Survey Administered Fall, 2012 (and again spring 2013)
When the class was complete, all students were asked to complete a survey giving their
impressions of the overall exercise and linking the whole exercise back to NASPAA’s five core
competencies. That survey is reproduced in this appendix.
31® Intemational System Dynamics Conference, Cambridge, MA (2013) 20
