Operationalizing Political Analysis in the East Asia Pacific Region

1
Operationalizing Political Analysis in the East
Asia Pacific Region

• An Agent Based Stakeholder Model

Please select a topic
Agenda
Example
Approach
Process
ABSM
Advice
Value
2
agenda

• Background on Operationalizing Political Analysis
• ABSM Approach Process
• A (Thinly) Disguised Example Demo
• Really Early Lessons

3
Political Analysis Tools
4
EAPS Political Analysis Quest A Search for
Relevance and Rigor

• (Amazing) Emerging Consensus that Politics Matter
  to Developmental Outcomes
• Growing Body of Analytic Work on Political
  Economy Issues
• Soft Tools for Incorporating Stakeholder Views in
  Program Design
• Interesting but Not Used
• Accessible but Not Rigorous
• Challenge How to develop operationally relevant
  and methodologically robust analytic tools that
  will change decisions on the ground in real time

5
From Early Pilot to New Improved Senturion

• Early Pilot with Agent-Based Model in 2 EAP
  Countries
• Consultants Performed Analysis
• Outside of Operational Cycle
• Little Operational Impact
• New Modeling Tool (Senturion)
• Desktop Software for In-House Use
• Modeling Built-in to Program
• Training for Bank Staff
• Feeds into Operational Decision Making

6
What is the ABSM Approach?

• Agent Based Stakeholder Model (ABSM) draws upon
  leading-edge work to analyze the preferences
  behaviors of relevant decision makers on
  politically driven issues.
• Provides (much) more systematic, dynamic
  modeling of complex interactions than traditional
  stakeholder analysis
• Employs microeconomics political (rational
  choice) theory algorithms to simulate bargaining
  dynamics coalition formation over time on
  specific issues.
• Accurate analysis can be generated using subject
  matter experts (not primary sources)
• Gauges in advance whether proposed policies are
  politically feasible as designed.
• Allows reform packages to be tailored to domestic
  political conditions.

7
How Does the ABSM Process Work?

• SENTURION defines problems as issues on which
  stakeholders compete to influence.
• With Subject Matter Experts (SMEs), SENTURION
  maps the current political landscape of
  stakeholders coalitions on issues.
• Power or Potential to Influence
• Position
• Importance or Salience
• The SENTURION ABSM then
• Tracks the complex interactions of all
  stakeholders using microeconomic political
  theory.
• Anticipates bargaining dynamics coalition
  formation over time.
• Identifies tactics to achieve desired outcomes.
• Allows comparison of alternative starting
  assumptions, hypotheses scenarios.

8
ABSM Process

• Framing
• Definition of Problem
• Structuring of Issues

Model Algorithms
Issue n
Knowledge Outcome Matrix
Issue 2
Issue1
Results
Continuum
Stakeholders
Issue Outcomes
Output

• Influence
• Position
• Importance

Paths
Data Collection
Consensus
Action Plan
Analysis
9
Initial Data
Creating a snapshot of the current political
landscape with experts, not to predict outcomes
but to render the context.

• Issue continuum
• List of stakeholders
• Position they adopt
• Potential to influence
• Salience of the issue

Where are politically powerful groups?
Where is compromise possible?
Where are stable coalitions?
Issue Continuum
10
Data Collection and Framing Process

• Interview of Experts
• Pair-wise Comparisons to Check Validity
• Data Refined in Framing Process (Issue continuum,
  Relative Positions, Power, Salience for
  Individual Actors and Groups -- with relative
  weights) are Checked, Re-checked, and Revised as
  Needed.

11
Conceptual Underpinnings of ABSM

• Median Voter Theory
• Game Theory

12
Median
The Median position for each party wins the most
votes
Number of American Voters
Republican Winner
Democratic Winner
R3
R2
R1
D3
D2
D1
Right
Left
Center
Political Orientation
13
Median
General Election The Race to the Center The
Median Position is the winning position in a
pairwise comparison of all alternatives (Duncan
Black 1958).
R2
D2
Number of American Voters
Right
Left
Center
Political Orientation
14
Risk the Median

• Risk Acceptant stakeholders are further from
  median or preferred outcome
• Risk Averse stakeholders are close to median or
  preferred outcome.

Deal makers
Freedom Fighters
Terrorists
Median
Extreme
Extreme
15
Risk Maximization
Every stakeholder wants to achieve their specific
goal, subject to risk tradeoffs
Less likely to reach a deal given more of what
you want
A

• Risk Acceptant Overvalue benefits of what
  stakeholder wants, undervalue costs of action.
• Risk Averse Overvalue costs of action,
  undervalue benefits of what stakeholder wants.
• Risk Neutral Similar evaluation of costs
  benefits.

B
Winning Personally
Get less of what you want for reaching a deal
C
Winning Politically
16
Game theory
A
wins
does
C
not
intervene
Maps all potential interactions and decision
calculations among all stakeholders based on how
risk colors perceptions
wins
B
A

C
3
rd
Party
win
Intervention
C
assists
A

A
C
lose
B
resists
enters
C
Challenge
dispute
Stakeholder
B

B
C
win
C
gives
B
assists
in
Stakeholder
wins
A
B
B

C
Decision?
A
Bilateral
lose
A
v B
wins
B
Status
Quo
Do Not
Challenge
D
Positive
D
In
Status
Quo
Negative
D
17
Game theoretic interactions

• The exact same process is repeated for all
  pairs of stakeholders
• This creates a social network of political
  relationships

A v. B value for action
B v. A value for action
?
B
A
18
Perceptual mapping

• Combining both As perceptions Bs
  perceptions about each other, we can derive their
  anticipated interaction
• Here A Bs anticipated interaction is
  conflictual

B
- Conflict
B
Advantage
A v. B value for action
Conflict
A
Advantage
- A
A
B v. A value for action
- B
19
Perceptual mapping
B

• Any potential combination of vectors is possible
  for any pairing
• The angle of the vector determines the
  relationship
• The length of the vector determines the intensity

- Bargain
- Conflict
Negotiate Compromise
B
Advantage
A v B
To
B
Conflict
- Impose
Yield to
B
A
Advantage
- A
A
Bargain
Negotiate Compromise
- Status Quo
To
A
Advantage
A v B
A v B
Impose
Status Quo
B
gives in to
A
- B
20
Network of proposals
As offer to B
A
B
Bs offer to A

• What are anticipated interactions?
• Offers
• Pressure
• Moves
• Leverage

As offer to C
As offer to D
Bs offer to C
C
D
21
2. Agent Based Rules Modeling Elements
1. Initial Stakeholder Data

• What is winning coalition or Median position?
• Given the winning coalition position, which
  groups are risk taking?
• How does each stakeholder view every other
  stakeholder on assisting or opposing the issue?
• Which stakeholders will make what proposals to
  other stakeholders, strengthening or weakening
  coalitions?
• Which stakeholders will revise their position on
  the issue allowing us to anticipate the
  political dynamics?

4. Modeling Iterations
3. Intermediate Dynamics

• How did Median position change?
• How did risk profiles change given the change in
  the median position?
• How do stakeholder perceptions change?
• Which stakeholders will make what proposals to
  other stakeholders given these changes?
• Which stakeholders will revise their position on
  the reform issue showing how the political
  dynamics might be modified?

5. Anticipated Outcome
6. Interpreting Outcomes
High

• Iterations stop when stakeholders see no further
  gains in discussions.
• Where key stakeholders end up on the issue
  determines the anticipated issue outcome.
• If a majority of stakeholders coalesce around a
  position, there is a large degree of consensus.
  If not, conflict will occur.

(InfluenceImportance)
Effective Power
Low
22
Early Experience in EAP Country X

• Very Problematic Governance Context
• High Risk Adjustment Operation (PRSO) Over a
  One-Year Period
• Which Reforms (nature, level, and extent) Should
  be Included as Prior Actions?
• How Likely are Key Stakeholders to Support
  Reforms (Both Initial Agreement and
  Implementation)

23
Country-Specific Problems

• Highly Sensitive Environment Premium on
  Confidentiality
• Limited Interviews Super-friendlies and Bank
  Staff
• Shifting Political Environment Adjustment in
  Data Inputs
• Changing Bank-Country Relationship More Broadly
  Possible Need to Consider New Issue (HR?)

24
Data Sourcing

• Process begins with Bank sources to minimize
  footprint
• IDs gaps in information and then go out to next
  sphere where necessary

Bank
Donors
NGOs Civil Society
Stakeholders
25
Suggested PRSO Reform Areas
NRM State Land Mapping Disclosure Centralized v.
Localized Forestry
PSD Single Window Risk Management Strategy
PFM Treasury Customs Civil Service Pay
PRSO-1 Risks
26
Public Finance Management
Treasury Cash
0 cash 100 checks and transfers
100 cash 0 checks s and transfers
0

100
Customs
Check payments 0
Check payments100
0

100
27
Public Finance Management
Civil Service Pay
No more increase across the board- All increase
thru MBPI
10-15 increase across the board
0

100
Donor Plan
100 Donor supplements pooled to support wage
bill
100 Donor project-based separate salary
supplements
0

100
28
PFM Treasury Treasury Cash Basecase Initial
Positions
29
PFM Treasury Cash Mitigation Strategy Good
cop/ Bad cop
30
PFM Treasury Treasury Cash Basecase Endgame
31
(No Transcript)
32
Operational Steps in ABSM Application

• Train Staff in ABSM Techniques
• Through Training and Pre-Mission Discussion,
  Begin Framing Issues
• Assure Comfort Level of CMU (Agree Protocol for
  mitigating potential security risks while
  collecting high quality data)
• Data Collection Interview Strategies
• PRSO Areas Defined
• Mission Work and Analysis
• Refinement of Prior Actions
• Brainstorming Discussion of Findings Working
  Through Tactics and Strategy
• Decision on PRSO
• Evaluation

33
Early Lessons

• Training is Intensive Profile of Required Staff
  Skills?
• Training Continues in Field Consultant
  Hand-Holding Needed
• Agree on Rules of Game w/ CMU (Client)
• Framing and Diligence Actually Changes Bank
  View/Understanding of Reform Definition and
  Requirements

34
Early Lessons (2)

• Relevance to Operational Needs Essential
• Availability to CMU for Ongoing Analysis (1 year
  license)
• To Mainstream, Business Model Has to Work
• Cost to CMU
• and Location of Trained Staff and Software

35
How to Evaluate Utility of ABSM?

• User Views (Mini-Survey)
• Degree of Use
• Change in Understanding of Issue
• Change in Approach
• Results on the Ground
• Disaster Averted
• Support for Reform Maximized

36
End Game?

• ABSM Mainstreamed (as one of Various Political
  Analysis Tools) in Bank for Routine and Special
  Tasks
• CAS Upstream
• Lending Operations
• Blue Sky
• Crisis Knowledge