Introduction Theory Design Results The Distortionary Effects of Classifications on Aid Allocation Decisions Lindsay Dolan Columbia University [email protected]November 18, 2017 Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
29
Embed
The Distortionary Effects of Classifications on Aid ... · Lindsay Dolan Columbia University [email protected] November 18, 2017 ... Bhutan Morocco Jordan Tuvalu Colombia
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Introduction Theory Design Results
The Distortionary Effects of Classifications onAid Allocation Decisions
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Motivation
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Dissertation Project
Classifications coordinate the perceptions and behaviors ofinfluential actors in the global economy.
I Evidence? Cross-national statistical analysis and interviewsshow that MICs receive less aid but are rated as moredemocratic and creditworthy than LICs
I Why? Classifications (1) simplify cognitive decision-makingand (2) allow certain actors to strategically justify theirdecisions
I So what?I Helps us understand power of IOs (Barnett and Finnemore 1999, 2004)
I Illustrates why beliefs and ideas affect market behavior (Blyth
2002; MacKenzie 2006; Shiller 2017)
I Fits into broader literature on labels, metrics, indicators ingovernance (Buthe 2012; Cooley and Snyder 2015; Gray 2013; Brooks, Cunha and Mosley
2015; Kelley and Simmons 2015)
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Dissertation Project
Classifications coordinate the perceptions and behaviors ofinfluential actors in the global economy.
I Evidence? Cross-national statistical analysis and interviewsshow that MICs receive less aid but are rated as moredemocratic and creditworthy than LICs
I Why? Classifications (1) simplify cognitive decision-makingand (2) allow certain actors to strategically justify theirdecisions
I So what?I Helps us understand power of IOs (Barnett and Finnemore 1999, 2004)
I Illustrates why beliefs and ideas affect market behavior (Blyth
2002; MacKenzie 2006; Shiller 2017)
I Fits into broader literature on labels, metrics, indicators ingovernance (Buthe 2012; Cooley and Snyder 2015; Gray 2013; Brooks, Cunha and Mosley
2015; Kelley and Simmons 2015)
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Explaining Aid Allocation
Why would donors use classifications to allocate aid?
I Puzzle: Classifications do not convey any private or expertinformation
1. CognitiveI Classifications simplify or distill information (Kahneman 2011)
I Conflicting information about developing countries (Jerven 2013)
2. StrategicI Using classifications created by credible third party improves
appearance of impartialityI “When we can show that most of our aid is benefiting Least
Developed Countries, this helps our numbers.”— U.S. Treasury official
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Allocation Activity
Timor-Leste $Zambia $South Africa $Tonga $El Salvador $
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Allocation Activity: Classifications Treatment
Timor-Leste $Zambia $South Africa $Tonga $El Salvador $
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Identifying the Classification EffectClassifications
Treatment Control
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Distinguishing the Mechanism
I Classifications treatment captures cognitive effect becausetreatment and control groups face identical incentives
I Introduce second treatment (factorial design) to vary thestrategic incentives participants face
I Control: Randomly selected participants receive prize
I Treatment: Randomly selected participants receive prize onlyif a “judge” approves of their allocations — “The judge willbe an intern (i.e an entry-level staffer) in the office of apolitician (i.e. a congressperson/MP) in a major donorcountry. The judge is NOT an expert on development.”
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Factorial DesignClassifications
Treatment Control
Treatment
Jud
ges
Control
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Identification Strategy
I How to operationalize a participant’s allocation?
I Ideal : Amount given to country if classified as lower-middleincome versus amount given to country if classified asupper-middle income
I Obstacles: Country-specific characteristics, no deceptionI Solution: Exploit variation in data
I GNI figures revised over timeI Identify instances when various estimates put above/below
thresholdI Circulate both versions (to different participants)I Compare allocations made by participants seeing “just-below”
version to those seeing “just-above” version
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Example
I Paraguay’s 2013 GNI per capita, reported in WDI inNovember 2014: $4,040
I Paraguay’s 2013 GNI per capita, reported in WDI inSeptember 2015: $4,190
I 2013 threshold separating LMICs from UMICs: $4,125
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Example
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
“Just-Below” RandomizationClassifications
Treatment Control
Treatment •◦•◦•
•◦•◦
◦•◦••
◦•◦•
◦◦•••
◦•◦◦
◦•◦◦◦
•◦••
Jud
ges
••◦◦◦
•◦◦•
◦•◦••
◦◦•◦
◦•••◦
•◦••
◦•◦◦•
•◦••
Control •◦◦•◦
◦◦••
•◦•◦•
◦◦•◦
•••◦◦
•◦◦◦
•◦•◦•
◦•◦◦
◦•◦◦•
◦•◦•
◦•◦••
◦••◦
◦◦•••
••◦◦
•◦••◦
◦•◦•
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Identifying Equation
Yij =α + γCi + λJBj + σJi+
δ1(Ci ∗ JBj) + δ2(Ji ∗ JBj) + δ3(Ci ∗ Ji )+
δ4(Ci ∗Mi ∗ JBj) + µj + ρj + εij
δ1 = Total effect of classificationsδ4 = Strategic effect of classificationsδ1 − δ4 = Cognitive effect of classificationsµr = Round F.E.ρr = Country identity F.E.Standard errors clustered at participant level
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Sample
I Fielded September 2017 at International Conference forSustainable Development (side meeting of UN GeneralAssembly)
I Fielded online and in person
I 232 participants (2088 rounds)
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Effects on Allocation to Threshold Country
treat_judge: 0 treat_judge: 1
treat_class: 0treat_class: 1
0 25 50 75 100 0 25 50 75 100
0.00
0.01
0.02
0.03
0.04
0.00
0.01
0.02
0.03
0.04
Allocation to threshold country
Den
sity
Just Above Just Below
Development Classifications and Aid Allocation Lindsay Dolan (Columbia University)
Introduction Theory Design Results
Effects on Allocation to Threshold Country
Allocation to threshold country (logged)TreatClass*TreatJudge*JustBelow (δ4) 0.269 0.158Strategic effect (0.174) (0.161)