401.661 Advanced Construction Technology
1
Moonseo Park
39동 433 Phone 880-5848, Fax 871-5518 E-mail: mspark@snu.ac.kr Department of Architecture College of Engineering Seoul National University Professor, PhD
Decision Making
Modeling human behaviors/Bounded rationality/Drug War
401.661 Advanced Construction Technology
401.661 Advanced Construction Technology
2
§ Must capture decision making as it is, NOT as it should be.
§ Rationality of human decision making is bounded.
§ Bounded rationality results from limitations on our knowledge, cognitive capabilities, and time.
§ Cognitive limitations: perception of information is not comprehensive but selective” (Hogarth 1987)
Bounded Rationality
401.661 Advanced Construction Technology
3
People and organizations have developed a number of ways to simplify the task of decision making.
§ Habit, Routines, and Rules of Thumb
§ Goal Formation and Satisficing
§ Problem Decomposition
Individual & Organizational
Responses to Bounded Rationality
401.661 Advanced Construction Technology
4
§ Decision makers are irrational or just plain stupid?
§ NOT at all
§ Human behavior is usually purposeful and motivated by a certain logic.
§ A decision rule is intendedly (=locally) rational if it would produce reasonable and sensible results if the actual
environment were as simple as the decision maker presumes it to be.
Intended Rationality
401.661 Advanced Construction Technology
5
§ An intendedly rational pricing policy can lead to an inadvertent price war.
§ Top: Mental model of a firm in which competitor prices are
believed to be exogenous. Cutting prices to Fill the Line ( B l ) when capacity utilization falls is locally rational if the boundary of
management’s mental model cuts the feedbacks to competitor prices.
§ Bottom: When competitor firms behave the same way and also cut prices to boost their own capacity utilization (B2), then the intendedly rational decision to lower prices in the hope of stimulating demand creates the reinforcing feedback R1 (shown by the thick lines) and a price war ensues whenever industry demand drops below capacity.
604 Part IV Tools for Modeling Dynamic Systems
F
IGURE 15-1Top: Mental model of a firm in which competitor prices are believed to be exogenous. Cutting prices to Fill the Line (Bl) when capacity utilization falls is locally rational if the boundary of management’s mental model cuts the feedbacks to competitor prices.
Bottom: When competitor firms behave the same way and also cut prices to boost their own capacity utilization (B2), then the intendedly rational decision to lower prices in the hope of stimulating demand creates the reinforcing feedback R1 (shown by the thick lines) and a price war ensues whenever industry demand drops below capacity. For clarity, additional feedbacks from prices to industry demand and from demand to capacity are not shown.
An intendedly rational pricing policy can lead to an inadvertent price war.
Management’s Mental Model: Competitor Price Is Exogenous
/ F
Demand\+
/CapacitjCapacity
til
Utilization Market Fill the LineShare
\
~ Competitor IndustryDemand
Actual Situation: Competitor Price Endogenous
The account above does assume the managers explicitly consider competitor prices but decide that the competitors won’t respond to a price cut with a price cut of their own. Alternatively, the managers may never consider competitor prices at all, assuming implicitly and without reflection that lower prices stimulate demand and will help fill the line.
401.661 Advanced Construction Technology
6
§ Only a very small number seem able to grow rapidly and steadily
§ No obvious differences between the successes and failures in fundamentals found
§ May caused by the different decision rules used to manage the enterprise and the unanticipated side effects of
policies that appeared to be rational and well-intentioned when viewed in isolation
Case Study: Modeling High-Tech Growth Firms
401.661 Advanced Construction Technology
7
§ Managers operating in the individual subunits are not assumed to understand the overall feedback structure.
§ For example, the capacity expansion decision is
based on product
availability as measured by the delivery delay and does not depend on
forecasts of future sales generated by the
marketing organization, which were generally distrusted and ignored by senior management.
606
f \
\ J
,Customer
Sales Contacts
Revenue
Delivery Delay Order
Demand
J
Capacity
/ L
FulfillmentJ
Orders
( @
Utilization/
)De> DelayPart IV Tools for Modeling Dynamic Systems
quite general and its lessons apply to growing organizations in any industry. The parameters are therefore chosen to be representative of typical high-tech products and not to replicate the experience of any one company.
To illustrate how interactions of intendedly rational policies could produce failure, Forrester deliberately made the strong simplifying assumption that the market for the firm’s product was unlimited. The potential of the computer and high-tech industry in the 1960s seemed to him to be so great that this was a rea- sonable assumption.
15.5.1 Model Structure: Overview
The model represents a single firm competing in a potentially unlimited market. To keep the model as simple as possible, Forrester deliberately omitted many organi- zational functions and structures. For example, there is no income statement or bal- ance sheet, and competitors are included implicitly in a simple market sector. The representation of the firm itself consists of three sectors, each representing a dif- ferent organizational subunit (Figure 15-2): sales, order fulfillment, and capacity acquisition.
FIGURE 15-2 Sectors of the market growth model
The model divides the firm into distinct organizational subunits. Each function operates on the basis of different information.
Capacity Acquisition
The Firm
I
The MarketSource: Adapted from Morecroft (1 983).
Sectors of the market growth model
The model divides the firm into distinct organizational subunits. Each
function operates on the basis of different information.
401.661 Advanced Construction Technology
8
Order Fulfillment
Variables outside the boundary are determined in other subsystems. Often the members of the subunit view these inputs as exogenous givens. Here, production capacity is taken to be outside the control of the order
fulfillment organization
Chapter 15 Modeling Human Behavior: Bounded Rationality or Rational Expectations?
F
Normal Delivery
i/ Delay
Desired Production
Order
Fulfillment
) +
Boo k-to-E3i I I Ratio
Capacity Utilization Rate
\
607
Interactions of organizational functions create important feedback loops, such as the balancing loop B3, which couples the firm to the market through product availability. An increase in orders boosts the backlog and delivery delay, causing some delivery-sensitive customers to take their business to the firm’s competitors.
The model represents the firm and its market as an ecology of interacting agents- the individual organizational functions-each with their own goals and decision rules.
Consistent with the principles of bounded rationality, managers operating in the individual subunits are not assumed to understand the overall feedback struc- ture. Each subunit is assumed to have access to and use a small number of infor- mational cues, not the full set of information potentially available. For example, the capacity expansion decision is based on product availability as measured by the delivery delay and does not depend on forecasts of future sales generated by the marketing organization, which were generally distrusted and ignored by senior management.
15.5.2 Order Fulfillment
Figure 15-3 shows the structure of the order fulfillment function.
F
IGURE15-3 Order fulfillment
The recta.ngle with rounded corners denotes the boundary of a subsystem or organizational subunit in
the model (Morecroft
1982).Variables outside the boundary are determined in other subsystems. Often
the members of the subunit view these inputs as exogenous givens. Here, production capacity is taken
tobe outside
thecontrol of the order fulfillment organization.
401.661 Advanced Construction Technology
9
Capacity Acquisition
§ Senior managers tend to be reluctant to invest until there is clear evidence of need.
§ The only reliable evidence that we need more capacity comes when we start missing delivery dates
§
Chapter 15 Modeling Human Behavior: Bounded Rationality or Rational Expectations? 61 1
FIGURE 15-6 Capacity acquisition
The rectangles around Delivery Delay Perceived by Company and Capacity represent delays without showing ,their full stock and flow structure.
Time for Company to
Perceive Delivery Delay Order
Rate
+
DeliveryCompany
-f
Capacity Expansion
E 4
Company
-
Goal for DeliveryDelay Pressure to
\+
Expand Capacity
Capacity Goal
+
Desired
w
Capacity + b a c i Acquisition t y
Delay
1 Effect of
Expansion Pressure on
Desired Capacity
The desired capacity decision is modeled as an anchoring and adjustment process (section 13.2.10). Management forms desired capacity by anchoring on current capacity, then adjusting it up or down based on various pressures. Because cues such as sales forecasts are noisy, unreliable, and untrustworthy, the pressure to expand capacity derives from the firm’s perceived ability to deliver compared to its goal.
(15-8) Desired Capacity
= Capacity
*
Effect of Expansion Pressure on Desired Capacity Effect of Expansion Pressure on Desired Capacity= f(Pressure to Expand Capacity) (15-9)
Delivery Delay Perceived by Company Company Goal for Delivery Delay
Pressure to Expand Capacity = (15-10)
Time for Company to Perceive Delivery Delay
(15-11) Delivery Delay
Perceived by Company
401.661 Advanced Construction Technology
61 6 Part IV Tools for Modeling Dynamic Systems
F
IGURE 15-1 0 Structure of the sales organizationThe gray link from Backlog to Shipment Rate represents the order fulfillment process (Figure 15-3) and closes the Sales Growth loop R1.
Sales Force Net Hiring Rate Sales Force
Adjustment
Time Target
Sales
/” +FT
Cost per Sales Representative
Sales
A Budaet
Sales Growth
Fraction of
f
Revenue to Sales
Change in Revenue +
Recent 1y
Revenue
f
Reporting Delay
Sales Effectiveness
Revenue
’+\
PriceThe order rate depends on the number of sales representatives and their effec- tiveness as measured by orders booked per person per month:
Order Rate = Sales Force
*
Sales Effectiveness (15-12) Sales effectiveness depends on the number of customer calls each sales person can make per month (assumed to be constant) and the fraction of calls resulting in a sale (the closing rate), which depends on the attractiveness of the product in the marketplace (see the market subsystem).10
Sales Force
401.661 Advanced Construction Technology
11
Behavior of the Full System
FIGURE 15-14 Behavior of the full market growth model
The potential revenue curve shows what revenue would be if the firm always had enough capacity to fill orders on time.
401
Potential Revenue ;i
L2 0
0 4000
3000
5
g
2000.
.I-.-
S3 1000
0
1 2
0 2 4 4 8 Month 7 2 9 6 120
0 2 4 4 8 Month 7 2 9 6 120
0
0 2 4 4 8 Month 7 2 9 6 120
a
6
0 2 4 4 8 Month 7 2 9 6 1 2 0
1.25
1 .oo
0.75 0.50 0.25 0.00
5 0 0
400
IG
r tn 300:
(D200 'fJ-
8
100 3
0
1 . 5
0 . 5 0
622
FIGURE 15-14 Behavior of the full market growth model
The potential revenue curve shows what revenue would be if the firm always had enough capacity to fill orders on time.
401
Potential Revenue ;i
L2 0
0 4000
3000
5
g
2000.
.I-.- S
3 1000
0
1 2
0 2 4 4 8 Month 7 2 9 6 120
0 2 4 4 8 Month 7 2 9 6 120
0
0 2 4 4 8 Month 7 2 9 6 120
a
6
0 2 4 4 8 Month 7 2 9 6 1 2 0
1.25
1 .oo
0.75 0.50 0.25 0.00
5 0 0 400
IG
r300 tn
:
(D200 'fJ-
8
100 3
0
1 . 5
0 . 5 0
622
*The potential revenue curve shows what revenue would be if the firm always had enough capacity to
fill orders on time.
401.661 Advanced Construction Technology
12
§ Overconfidence: the confidence bounds people provide around their estimate of an unknown quantity are too narrow.
§ More information, more confident, while accuracy did not improve.
§ Thousands of repetitions provide feedbacks enabling to learn from experience (weather forecast, gambling) but there is
little chance to learn from experience in most social/business situations.
Examples:
¨
The Challenger explosion estimated 1 in 100,000.
¨
Underestimating the likelihood of declining share price during 1920s and 1980s-90s
Overconfidence
401.661 Advanced Construction Technology
13
§ List all the reasons your opinion could be wrong.
§ Solicit the opinions of a diverse group especially those with opposite views.
§ Suspect statements that something is absolutely certain, inevitable or a one in million chance.
§ When formal models are available, conduct extensive sensitivity tests.
Overcoming Overconfidence
401.661 Advanced Construction Technology
14
The War on Drugs
§ Use of Cocaine dramatically increased in 1980s
§ Billions spent to increase enforcement, focusing on the supply side
§ On demand side: “Just say NO”
§ Did it work?
401.661 Advanced Construction Technology
15
It seemed to be working…
Down from 3% in 1985 to 1% in 1990
0 0.01 0.02 0.03 0.04
1976 1978 1980 1982 1984 1986 1988 1990
Prevalence (fraction of population) NHS Past Month
User Fraction
401.661 Advanced Construction Technology
16
But, the problem was getting worse…
0 200 400 600 800
1976 1978 1980 1982 1984 1986 1988 1990
Price (1990 $/gram)
Pure
Adulterated 0
200 400 600
1976 1978 1980 1982 1984 1986 1988 1990
Thousand people/year
Total Arrests
Sale/Manufacture Possession
0 20 40 60
1976 1978 1980 1982 1984 1986 1988 1990
Thousand people/year
ER Mentions
ME Mentions * 10
0 20 40 60 80 100
1976 1978 1980 1982 1984 1986 1988 1990
Purity (%)
Source: Homer (1993, 1997).
401.661 Advanced Construction Technology
17
As a result,..
§ Cocaine use was up sharply and availability was growing.
§ The same failure of prohibition in 1920s and 1930s.
§ Critics argue “ interdiction could never work
and call for stronger demand-side measure
(MacCoun and Reuter, 1997)”
401.661 Advanced Construction Technology
18
How to explain?
Supporters Critics
• Rising arrest rates by
greater enforcement, not by greater drug use
• Falling prices, rising purity, surge in ER by substitution of more potent crack for the less pure power form
• Cocaine users are less likely than law-abiding citizens to be selected for the survey.
Thus,
• They are likely deny they
use drugs.
401.661 Advanced Construction Technology
19
System Dynamics Model
The national Institute of Justice commissioned a study, 1980s
Source: Homer (1993, 1997).
Death Rate (All Causes) Transitional
User Relapse Rate Escalation to
Compulsive Use
Active Compulsive
Users
Total Active Users
Ex-Users
Ex-User Relapse
Rate
Quit Rate Initiation to
Cocaine Use Population
Increase
Death Rate (All Causes) Never Used
Population
Ever-Used Population Transitional
Users Cessation
Rate
Active
Casual
Users
401.661 Advanced Construction Technology
20 Supply
Disruption
Fear of Arrest B4
Negative Health and Social Effects
B3
Scale and Learning
Effects R2 Word of Mouth
R1
Cocaine- Related
Crime
Seizure Fraction Reported
Prevalence of Cocaine Use
Enforcement Intensity;
Sentence Severity Social
Acceptability of Drugs
Cocaine-Related Arrests and Incarceration Perceived
Legal Risks
Social Exposure to
Cocaine
- -
-
-
+ +
-
+
+ +
+
+ +
+ +
-
- +
+ +
Delay
Delay
Delay
Cocaine Market
• Retail price
• Purity
• Retail sales
• Imports
• Consumption +
-
Clean up the Streets
B2
Underreporting of Cocaine Use
(Fraction of Users Lying)
Perceived Health
Risks
Cocaine User Population Disaggregated by
• Preferred form
• Intensity of use
• Recency of use
B1
Morbidity and Mortality
Source: Homer (1993, 1997).
401.661 Advanced Construction Technology
21
Simulated / Actual Data
0 200 400 600 800
1976 1980 1984 1988 1992 1996
Thousand people/year
Sale/Mfr
Possession
Total Arrests
History Forecast
401.661 Advanced Construction Technology
22
0 25 50 75
1976 1980 1984 1988 1992 1996
Thousand people/year
ER Mentions
ME Mentions * 10
History Forecast
401.661 Advanced Construction Technology
23
0 0.01 0.02 0.03 0.04 0.05
1976 1980 1984 1988 1992 1996
Prevalence (fraction of population)
NHS Past Month User Fraction Simulated
Actual Past Month User Fraction
Simulated
Reported Past Month User Fraction
History Forecast
401.661 Advanced Construction Technology
24
Where did they go?
401.661 Advanced Construction Technology
25
In 2004, The South Korean government passed an anti-prostitution law (Special Law on Sex Trade 2004) prohibiting the buying and selling of sex and shutting down brothels.
In 2007, the courts prosecuted 35,000 clients, 2.5 times higher than the number of those who were caught buying sex in 2003.
Meanwhile enforcement is weak and corruption problematic; there is little evidence that new legislation has made much difference, the trade simply finding other ways to carry on its business.
Despite legal sanctions and police crackdowns, prostitution continues to flourish in S Korea, while sex workers continue to actively resist the
state's activities.
401.661 Advanced Construction Technology
26
Identify S&F….
401.661 Advanced Construction Technology
27
A8: Suggest Policy Alternatives
§ Build CLDs
§ Combine them with S&F
§ Test behaviors of your model with the empirical data
§ Suggest policy alternatives based on the model behaviors
§ Due next week
* Do not have to do site survey.
401.661 Advanced Construction Technology
28