1 Principals and Interests: Collective Principals and Environmental Lending at Multilateral Development Banks Daniel L. Nielson Brigham Young University.

1

Principals and Interests: Collective Principals and Environmental Lending at

Multilateral Development Banks

Daniel L. NielsonBrigham Young University

Michael J. TierneyCollege of William and Mary

2

Empirical Puzzle

Trends Member governments grow more environmentalist after

early-1970s MDBs largely ignore environment until late-1980s Late-1980s through 1990s – Big increase in MDB

environmental lending

Gaps How can we explain delay and eventual adoption of an

environmental agenda? How can we explain timing of adoption across different

MDBs?

3

IR Theory and IOs

Neorealism and Neoliberalism deny IO agency

Constructivism suggests abdication Agency theory resolves gaps

IOs are independent actors Member states conceived as principals of IOs Our distinction: collective principal

Principals’ converging preferences guide agents Principals’ coordination problems enable agency slack

4

Modeling Collective Delegation

Which type of principal?

Single Principal

Collective Principal

Multiple Principals

Most IOs

X Agent XYZ Agent Y Agent

Z

X

5

Two Stages of Delegation at MDBs: Asian Development Bank

US

Japan

Laos

Afghanistan

Bhutan

A

B

C

ASDB

States Executive

Board MDB

6

Collective vs. Multiple Principals, Stage 1

0% 50% 100% 25% 75%

X Y Z

X

Collective Principal with Majority Vote

Proportion of Environmental Loans

Policy Outcome

0% 50% 100%25% 75%

X

Y

Z

X

Proportion of Environmental Loans

Multiple Principals, Independent Action

Observational Equivalence

Policy Outcome

7

Collective vs. Multiple Principals, Stage 2

0% 50% 100% 25% 75%

XYZ Collective Principal

with Majority Vote

Proportion of Environmental Loans

X

Policy Outcome

Divergent Expectations

0% 50% 100%25% 75%

X

YZ

Proportion of Environmental Loans

Multiple Principals, Independent Action

Policy Outcome?

8

What Causes IO Agents to Change Behavior? Hypothesis: As the policy preferences of collective

principals shift toward environmental concerns MDB Environmental loans will increase MDB Neutral loans will increase MDB Dirty loans will decrease

9

Imputing & Aggregating Preferences (IVs) Imputing Preferences

Infer preferences from behavior – revealed preferences

Use policy outcomes as proxies for preferences Environmental Policy Index Environmental Foreign Aid

Aggregating Preferences Examine states’ preference distribution Predict voting coalitions Weight state’s influence by degree it proves

“pivotal” to potential coalitions

10

Data on Dependent Variable More than 7,500 individual development loans

World Bank (IBRD & IDA) African Development Bank & Fund Asian Development Bank Inter-American Development Bank Islamic Development

Coded all loans on five-point environmental scale Dirty Strictly Defined: direct negative impact (i.e., logging) Dirty Broadly Defined: moderate but negative (agriculture) Neutral: no immediate impact (education, telecomm.) Environmental Broadly Defined: preventative (nuclear safety) Environmental Strictly Defined: direct (pollution control,

biodiversity protection)

11

MDB Environmental Lending

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

Year

Per

cen

t E

nvi

ron

men

tal

& N

eutr

al P

roje

cts

AFDFIADBIBRDIDAISDB

12

Environmental Policy Preferences - Policy Index

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

11

98

0

19

81

19

82

19

83

19

84

19

85

19

86

19

87

19

88

19

89

19

90

19

91

19

92

19

93

19

94

19

95

19

96

19

97

19

98

19

99

20

00

Year

En

viro

nm

enta

l P

refe

ren

ces

(Po

licy

In

dex

)

AFDBAFDFASDBIADBIBRDIDAISDB

13

Variables of Interest Dependent Variable: Environmental Impact

1=DSD, 2=DBD, 3=N, 4=EBD, 5=ESD

Key Indep. Variable: Environmental Preferences Measured 3 ways:

Environmental Policy Index Environmental Foreign Aid / Total Aid Environmental + Neutral Aid / Total Aid

Controls Organic Water Pollution, (De)forestation, Threatened Birds,

Sanitation, Infant Mortality, Fertility Rate, Agricultural Value Added, CITES Commitments, GDP Per Capita, ln(GDP), ln(Population), Domestic Savings, Exports, Vehicles, Protected Land

14

Ordered Logit Regression ResultsDep. Var.: Environmental Impact

Environmental Environmental Environmental Independent Variables Preferences: Preferences: Preferences: Policy Index Foreign Aid Foreign Aid

(w/Neutral) Environmental Policy Preferences

1.137***

(0.21) Environmental For. Aid Prefs. 2.771*** (0.20) Enviro. Aid Prefs. (incl. Neutral) 2.097*** (0.16) Fertility Rate -0.0787*** -0.0255 -0.0122 (0.029) (0.031) (0.030) Savings -0.00974*** -0.00596** -0.00544** (0.0030) (0.0024) (0.0026) GDP Per Capita (1k US $) 0.0602*** 0.0241 0.0397* (0.023) (0.020) (0.023) Agriculture Value Added -0.00726** -0.00538* -0.00411 (0.0037) (0.0032) (0.0031) Threatened Birds 0.268 1.028* 0.982* (0.64) (0.59) (0.57) Forestation (Deforestation) -1.042*** -0.886*** -0.823*** (0.19) (0.19) (0.19) Sanitation 0.00308** 0.00284** 0.00198 (0.0014) (0.0013) (0.0013) Protected Land 0.524 0.913** 0.819* (0.47) (0.44) (0.43) Number of Observations 7538 7538 7538 Variables tested but insignificant statistically: Log of GDP, Log of Population, CITES Commitments Kept, Protected Land, Organic Water Pollution, Vehicles, Exports / GDP, Infant Mortality Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

15

Results Summary Environmental Preferences:

Positive and Significant (beyond .001 level) Substantively Important

+1 stdev (.12 to .16) for Environmental Preferences 4.6% to 8.3% in probability of Dirty project 2.5% to 4.9% in probability of Neutral project 1.6% to 3.4% in probability of Environmental project

Controls: mixed to weak Only Savings and (De)Forestation performed

consistently Others significant in 2 specifications:

GDP per capita, Agriculture Value Added, Threatened Birds, Sanitation, and Protected Land

16

Conclusion

Collective delegation can work In (Most) Difficult Context

International Anarchy Extreme Preference Heterogeneity Many Actors (up to 180)

More work to Other specifications of preferences Further robustness checks

17

Extra Slide Pivotal-Weighted

Preferences

Pivotal-Weighted Preferences

ActorVote

Share Ideal Pivot.Ideal*Pivot.

A 0.2 2 0.2 0.4

B 0.3 5 0.4 2.0

C 0.1 6 0.2 1.2

D 0.3 8 0.2 1.6

E 0.1 9 0.0 0.0

Sum 5.2

Actors’ Ideal Points:

  A   B  C D  E  

1 2 3 4 5 6 7 8 9 10

Potential Pivotal Coalitions PlayersABC A or CABCD NoneABCDE NoneBCD B or DBCDE B