elite capture of foreign aid: evidence from offshore bank

70
Elite Capture of Foreign Aid: Evidence from Offshore Bank Accounts Jørgen Juel Andersen 1 Niels Johannesen 2 Bob Rijkers 3 1 BI Norwegian Business School, Oslo 2 University of Copenhagen 3 World Bank

Upload: others

Post on 17-Nov-2021

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Elite Capture of Foreign Aid:Evidence from Offshore Bank Accounts

Jørgen Juel Andersen1 Niels Johannesen2 Bob Rijkers3

1BI Norwegian Business School, Oslo

2University of Copenhagen

3World Bank

February 28, 2020

Page 2: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Introduction

I Classic concern: elite capture of aid

I Consistent withI theories of rent seeking (e.g. Svensson, 2000)I corruption in aid-receiving countries (Alesina and Weder,

2002)I Difficult to measure

I Do World Bank aid disbursements trigger increases inforeign deposits in bank accounts in havens?

Page 3: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Introduction

I Classic concern: elite capture of aidI Consistent with

I theories of rent seeking (e.g. Svensson, 2000)

I corruption in aid-receiving countries (Alesina and Weder,2002)

I Difficult to measure

I Do World Bank aid disbursements trigger increases inforeign deposits in bank accounts in havens?

Page 4: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Introduction

I Classic concern: elite capture of aidI Consistent with

I theories of rent seeking (e.g. Svensson, 2000)I corruption in aid-receiving countries (Alesina and Weder,

2002)

I Difficult to measure

I Do World Bank aid disbursements trigger increases inforeign deposits in bank accounts in havens?

Page 5: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Introduction

I Classic concern: elite capture of aidI Consistent with

I theories of rent seeking (e.g. Svensson, 2000)I corruption in aid-receiving countries (Alesina and Weder,

2002)I Difficult to measure

I Do World Bank aid disbursements trigger increases inforeign deposits in bank accounts in havens?

Page 6: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Introduction

I Classic concern: elite capture of aidI Consistent with

I theories of rent seeking (e.g. Svensson, 2000)I corruption in aid-receiving countries (Alesina and Weder,

2002)I Difficult to measure

I Do World Bank aid disbursements trigger increases inforeign deposits in bank accounts in havens?

Page 7: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Introduction (continued)

I Havens: offshore banking centers known for secrecyand asset protection havens

I Main sample: 22 highly aid dependent countries,1990-2010 highly aid dependent countries

I annual WBG aid>2% GDPI absorb 10% of all WB lendingI Extensions

I period since 2009I less aid dependent countries

Page 8: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Introduction (continued)

I Havens: offshore banking centers known for secrecyand asset protection havens

I Main sample: 22 highly aid dependent countries,1990-2010 highly aid dependent countries

I annual WBG aid>2% GDPI absorb 10% of all WB lendingI Extensions

I period since 2009I less aid dependent countries

Page 9: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Premise: Haven deposits belong to elites

I Offshore wealth is extremely concentratedI Scandinavia: Top 0.01% (0.1%) richest households own

50% (80%) of wealth on offshore accounts (Alstadsaeter etal., 2017)

I By contrast, the poor are mostly unbankedI 9% of the bottom 40% households had a (domestic) bank

account in 2011 (Demirguc-Kunt et al, 2011)

Page 10: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Premise: Haven deposits belong to elites

I Offshore wealth is extremely concentratedI Scandinavia: Top 0.01% (0.1%) richest households own

50% (80%) of wealth on offshore accounts (Alstadsaeter etal., 2017)

I By contrast, the poor are mostly unbankedI 9% of the bottom 40% households had a (domestic) bank

account in 2011 (Demirguc-Kunt et al, 2011)

Page 11: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Main findings

I WB aid disbursements to highly aid dependentcountries coincide with significant increases in havendeposits within the same quarter

I Aid-induced increase in haven deposits ("Leakage rate")I ≈ 7.5% aid inflows for the average countryI ≈ 5% of aggregate flows to these countriesI tends to increase with ratio of aid/GDP

Page 12: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Main findings

I WB aid disbursements to highly aid dependentcountries coincide with significant increases in havendeposits within the same quarterI Aid-induced increase in haven deposits ("Leakage rate")

I ≈ 7.5% aid inflows for the average countryI ≈ 5% of aggregate flows to these countriesI tends to increase with ratio of aid/GDP

Page 13: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Additional findings

I Disbursements do not trigger flows to non-havens

I Improved financial transparency since 2009 has nothelped

Page 14: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Additional findings

I Disbursements do not trigger flows to non-havens

I Improved financial transparency since 2009 has nothelped

Page 15: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Interpretation

I Isolating the underlying mechanism is challenging

I we don’t know who owns the deposits=> cannot attribute agency

I Elite capture is a salient and plausible explanation

I Other explanations are possible

We would love to hear your thoughts!

Page 16: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Interpretation

I Isolating the underlying mechanism is challengingI we don’t know who owns the deposits

=> cannot attribute agency

I Elite capture is a salient and plausible explanation

I Other explanations are possible

We would love to hear your thoughts!

Page 17: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Interpretation

I Isolating the underlying mechanism is challengingI we don’t know who owns the deposits

=> cannot attribute agency

I Elite capture is a salient and plausible explanation

I Other explanations are possible

We would love to hear your thoughts!

Page 18: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Interpretation

I Isolating the underlying mechanism is challengingI we don’t know who owns the deposits

=> cannot attribute agency

I Elite capture is a salient and plausible explanation

I Other explanations are possible

We would love to hear your thoughts!

Page 19: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Key Caveats & Common (Media) misperceptions (1/2)

I Correlation does not imply causationI money is fungible

Page 20: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Key Caveats & Common (Media) misperceptions (2/2)

I our main estimates only apply to 22 highly aiddependent countries

I these account for 10% of all WB lendingI aggregate leakage in these countries = 5%

Page 21: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Outline

Introduction

Data

Empirical Strategy

ResultsMain ResultsRobustnessHeterogeneityCalculating Leakage

Interpretation

Conclusions

Supplemental Slides

Page 22: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Cross border bank deposits

I Quarterly data on bilateral non-bank deposits, i.e.privately held accountsI only observe direct ownership

I Source: Locational Banking Statistics of the Bank forInternational Settlements (BIS)I covers value of bank deposits in 43 financial centers owned

by residents of around 200 countriesI confidential data: 1990-2010I public data with partial coverage: 1990-2018

I Havenit : sum of all deposits owned by residents ofcountry i in the 17 havens in quarter t

Page 23: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Foreign aid

I The World Bank Project Database (WBPD)I Approval dateI Commitment amountI SectorI Lending instrument type (e.g., investment projects or

development policy loans)I Timing of each disbursement

I Aidit : Sum of all IDA and IBRD project disbursementsin country i at time t (normalized by GDP)

Page 24: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Other variables

I Wars - PRIOI Coups - Powell and Thyne (2011)I Financial Crises - Laeven and Valencia (2012)I Natural Disasters - International Disaster DatabaseI Petroleum Rents - WDII Financial Sector Development - WDII Capital Account Openness - Chinn and Ito (2006)I Disclosure Requirements - Djankov et al. (2010)I Democracy - Polity IVI Corruption - WGII Procurement - World Bank Procurement Database

Page 25: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Aid and haven deposits growth distributions

050

100

150

200

250

Freq

uenc

y

0 .01 .02 .03 .04World Bank aid (relative to GDP)

Figure: Aid (% of GDP)0

100

200

300

400

Freq

uenc

y

-1 -.5 0 .5 1 1.5Change in haven deposits (in log-levels)

Figure: Haven deposits Growth

Page 26: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Descriptive Statistics

Variable Mean Std. Dev. Min MaxAidWB disbursements (% GDP) 0.027 0.006 0.020 0.039Total foreign aid (%GDP) 0.101 0.078 0.001 0.647Foreign depositsHaven deposits (million USD) 61 68 0 621Non-haven deposits (million USD) 136 143 0 964Quarterly growth ratesHaven deposits growth 0.020 0.272 -0.875 1.240Non-haven deposits growth 0.015 0.209 -0.693 0.795GDP growth 0.017 0.040 -0.240 0.178

crowding in

Page 27: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Empirical Strategy

∆ log(Havenit ) = βAidit + γXit + µi + τt + εit

I ∆ log(Havenit ) growth rate in haven deposits owned bycountry i in quarter t

I Aidit quarterly WB aid disbursements/GDPI Xit : ∆ log GDPit (in robustness: foreign exchange rate; oil

wealth shocks)I µi : country fixed effectsI τt : time fixed effectsI Standard errors clustered at the country levelI Deposit and aid variables winsorized at the 1%/99% level.I Alternative specifications/LHS-variables:

∆ log (nonhavenit ) and ∆ log (havenit )−∆ log (nonhavenit )

Page 28: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Empirical Strategy

∆ log(Havenit ) = βAidit + γXit + µi + τt + εit

I ∆ log(Havenit ) growth rate in haven deposits owned bycountry i in quarter t

I Aidit quarterly WB aid disbursements/GDP

I Xit : ∆ log GDPit (in robustness: foreign exchange rate; oilwealth shocks)

I µi : country fixed effectsI τt : time fixed effectsI Standard errors clustered at the country levelI Deposit and aid variables winsorized at the 1%/99% level.I Alternative specifications/LHS-variables:

∆ log (nonhavenit ) and ∆ log (havenit )−∆ log (nonhavenit )

Page 29: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Empirical Strategy

∆ log(Havenit ) = βAidit + γXit + µi + τt + εit

I ∆ log(Havenit ) growth rate in haven deposits owned bycountry i in quarter t

I Aidit quarterly WB aid disbursements/GDPI Xit : ∆ log GDPit (in robustness: foreign exchange rate; oil

wealth shocks)

I µi : country fixed effectsI τt : time fixed effectsI Standard errors clustered at the country levelI Deposit and aid variables winsorized at the 1%/99% level.I Alternative specifications/LHS-variables:

∆ log (nonhavenit ) and ∆ log (havenit )−∆ log (nonhavenit )

Page 30: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Empirical Strategy

∆ log(Havenit ) = βAidit + γXit + µi + τt + εit

I ∆ log(Havenit ) growth rate in haven deposits owned bycountry i in quarter t

I Aidit quarterly WB aid disbursements/GDPI Xit : ∆ log GDPit (in robustness: foreign exchange rate; oil

wealth shocks)I µi : country fixed effectsI τt : time fixed effects

I Standard errors clustered at the country levelI Deposit and aid variables winsorized at the 1%/99% level.I Alternative specifications/LHS-variables:

∆ log (nonhavenit ) and ∆ log (havenit )−∆ log (nonhavenit )

Page 31: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Empirical Strategy

∆ log(Havenit ) = βAidit + γXit + µi + τt + εit

I ∆ log(Havenit ) growth rate in haven deposits owned bycountry i in quarter t

I Aidit quarterly WB aid disbursements/GDPI Xit : ∆ log GDPit (in robustness: foreign exchange rate; oil

wealth shocks)I µi : country fixed effectsI τt : time fixed effectsI Standard errors clustered at the country level

I Deposit and aid variables winsorized at the 1%/99% level.I Alternative specifications/LHS-variables:

∆ log (nonhavenit ) and ∆ log (havenit )−∆ log (nonhavenit )

Page 32: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Empirical Strategy

∆ log(Havenit ) = βAidit + γXit + µi + τt + εit

I ∆ log(Havenit ) growth rate in haven deposits owned bycountry i in quarter t

I Aidit quarterly WB aid disbursements/GDPI Xit : ∆ log GDPit (in robustness: foreign exchange rate; oil

wealth shocks)I µi : country fixed effectsI τt : time fixed effectsI Standard errors clustered at the country levelI Deposit and aid variables winsorized at the 1%/99% level.

I Alternative specifications/LHS-variables:

∆ log (nonhavenit ) and ∆ log (havenit )−∆ log (nonhavenit )

Page 33: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Empirical Strategy

∆ log(Havenit ) = βAidit + γXit + µi + τt + εit

I ∆ log(Havenit ) growth rate in haven deposits owned bycountry i in quarter t

I Aidit quarterly WB aid disbursements/GDPI Xit : ∆ log GDPit (in robustness: foreign exchange rate; oil

wealth shocks)I µi : country fixed effectsI τt : time fixed effectsI Standard errors clustered at the country levelI Deposit and aid variables winsorized at the 1%/99% level.I Alternative specifications/LHS-variables:

∆ log (nonhavenit ) and ∆ log (havenit )−∆ log (nonhavenit )

Page 34: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Endogeneity

I Problem: Aid disbursements are endogenousI Natural disasters, financial crises and conflict can trigger

simultaneous surge in both aid and offshore deposits

I SolutionsI Use lags and leadsI Use an IV: Predicted disbursementsI Exclude periods of turmoil: war, coups, financial crises

Page 35: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Endogeneity

I Problem: Aid disbursements are endogenousI Natural disasters, financial crises and conflict can trigger

simultaneous surge in both aid and offshore deposits

I SolutionsI Use lags and leadsI Use an IV: Predicted disbursementsI Exclude periods of turmoil: war, coups, financial crises

Page 36: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Instrumenting Aid (Kraay, 2012, 2014)

-.01

0.0

1.0

2.0

3ai

d (re

sidu

aliz

ed)

-.01 0 .01 .02instrument (residualized)

coef = .87, se = .08

-.01

0.0

1.0

2.0

3ai

d (re

sidu

aliz

ed)

-.01 -.005 0 .005 .01instrument (residualized)

coef = .77, se = .09

I Predict quarterly disbursements for each project usingI initial commitments

I average disbursement schedule in the same sector andregion disbursementprofiles

I aggregate across all projects (excluding initial quarters)

Page 37: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Instrumenting Aid (Kraay, 2012, 2014)

-.01

0.0

1.0

2.0

3ai

d (re

sidu

aliz

ed)

-.01 0 .01 .02instrument (residualized)

coef = .87, se = .08

-.01

0.0

1.0

2.0

3ai

d (re

sidu

aliz

ed)

-.01 -.005 0 .005 .01instrument (residualized)

coef = .77, se = .09

I Predict quarterly disbursements for each project usingI initial commitmentsI average disbursement schedule in the same sector and

region disbursementprofiles

I aggregate across all projects (excluding initial quarters)

Page 38: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Instrumenting Aid (Kraay, 2012, 2014)

-.01

0.0

1.0

2.0

3ai

d (re

sidu

aliz

ed)

-.01 0 .01 .02instrument (residualized)

coef = .87, se = .08

-.01

0.0

1.0

2.0

3ai

d (re

sidu

aliz

ed)

-.01 -.005 0 .005 .01instrument (residualized)

coef = .77, se = .09

I Predict quarterly disbursements for each project usingI initial commitmentsI average disbursement schedule in the same sector and

region disbursementprofiles

I aggregate across all projects (excluding initial quarters)

Page 39: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Baseline Results

(1) (2) (3)haven non-haven ratio

(% change) (% change) (% change)

Aid (%GDP) 3.391*** -1.511 4.973***(1.154) (1.035) (1.715)

GDP (%growth) 0.122 0.147** -0.001(0.141) (0.063) (0.150)

country FE Yes Yes Yestime FE Yes Yes YesObservations 1,648 1,652 1,645R-squared 0.101 0.092 0.076Standard errors are clustered by country

Page 40: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Haven deposits increase when aid flows in

-50

5Pe

rcen

tage

cha

nge

in h

aven

dep

osits

-4 -2 0 2 4Quarter relative to disbursement

Page 41: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Non-haven deposits do not respond to disbursements

-50

5Pe

rcen

tage

cha

nge

in n

on-h

aven

dep

osits

-4 -2 0 2 4Quarter relative to disbursement

Page 42: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Aid differentially impacts haven vs non-haven deposits

-50

5Pe

rcen

tage

cha

nge

in h

aven

to n

on-h

aven

ratio

-4 -2 0 2 4Quarter relative to disbursement

Page 43: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Instrumenting Aid

Dependent variable: % Change in haven deposits(1) (2) (3)

OLS IV IVBaseline exclude 1 quart exclude 3 quart

Aid (%GDP) 3.391*** 2.595*** 2.848*(1.154) (1.004) (1.564)

country FE Yes Yes Yestime FE Yes Yes YesKleibergen-Paap rk Wald-F 182.624 159.650Observations 1,648 1,648 1,648R-squared 0.101 0.005 0.005Standard errors are clustered by country

Page 44: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Robustness

Dependent variable: % Change in haven depositsOmit periods with other shocks

No Wars No Coups No Disasters No Fin Crisis(1) (2) (3) (4)

Aid (%GDP) 2.754** 3.446*** 3.601*** 3.559***(1.296) (1.160) (1.237) (1.190)

Observations 1,386 1,618 1,586 1,511R-squared 0.107 0.100 0.104 0.102

Alternative specificationscountry-year FE Exchange controls Oil*time FE No winsorization

(5) (6) (7) (8)

Aid (%GDP) 3.290* 3.310*** 3.535** 3.591**(1.895) (1.126) (1.296) (1.500)

Observations 1,648 1,648 1,648 1,648R-squared 0.242 0.103 0.142 0.100

Standard errors are clustered by country. Country and time FE included

Page 45: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Heterogeneity C

orru

ptio

n C

redi

t D

emoc

racy

Dis

clos

ure

Pro

cure

men

t

low corruption controlhigh corruption control

disclosure requiredno disclosure required

less democraticmore democratic

low credithigh credit

high local procurementlow local procurement

-5 0 5 10

Estimated impact of aid on haven deposits

Page 46: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Does Type of Aid Matter?

(1) (2) (3)haven non-haven ratio

(% change) (% change) (% change)

IPF Aid (% of GDP) 5.122*** -0.435 5.787**(1.708) (1.563) (2.256)

DPF Aid (% of GDP) 2.697* -1.973 4.704**(1.348) (1.323) (2.060)

Observations 1,648 1,652 1,645R-squared 0.100 0.092 0.076

All specifications control for GDP growth and country and time FE.Standard errors are clustered by country. Country and time FE are included.

Page 47: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Transparency (post 2008) has not helped

-20

24

68

β (Ai

d/G

DP)

1990-2009 2010-2019

Source: Andersen, Johannesen, and Rijkers (2019)

Impact of aid on growth in offshore deposits (public data)

Page 48: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Quantifying the impact of aid on offshore depositsLeakage rate ≡ Aid-induced increase in offshore deposits

Calculating average leakageI βAid = 3.4

I => 1% Aid(%GDP) ↑ => 3.4% haven deposits ↑

I mean of haven/GDP: 0.022 (in the high aid sample)I haven/GDP increases by (3.4× 0.022 =) 0.075 when

Aid/GDP increases by 1=> leakage rate is 7.5%

Calculating aggregate leakageI Assume βAid = 3.4 applies uniformlyI mean of haven/GDP weighted by the share of average aid

received: 0.014I haven/GDP increases by (3.4× 0.014 =) 0.05 when

Aid/GDP increases by 1.=> leakage rate is 5%

Page 49: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Quantifying the impact of aid on offshore depositsLeakage rate ≡ Aid-induced increase in offshore deposits

Calculating average leakageI βAid = 3.4

I => 1% Aid(%GDP) ↑ => 3.4% haven deposits ↑I mean of haven/GDP: 0.022 (in the high aid sample)

I haven/GDP increases by (3.4× 0.022 =) 0.075 whenAid/GDP increases by 1=> leakage rate is 7.5%

Calculating aggregate leakageI Assume βAid = 3.4 applies uniformlyI mean of haven/GDP weighted by the share of average aid

received: 0.014I haven/GDP increases by (3.4× 0.014 =) 0.05 when

Aid/GDP increases by 1.=> leakage rate is 5%

Page 50: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Quantifying the impact of aid on offshore depositsLeakage rate ≡ Aid-induced increase in offshore deposits

Calculating average leakageI βAid = 3.4

I => 1% Aid(%GDP) ↑ => 3.4% haven deposits ↑I mean of haven/GDP: 0.022 (in the high aid sample)I haven/GDP increases by (3.4× 0.022 =) 0.075 when

Aid/GDP increases by 1=> leakage rate is 7.5%

Calculating aggregate leakageI Assume βAid = 3.4 applies uniformlyI mean of haven/GDP weighted by the share of average aid

received: 0.014I haven/GDP increases by (3.4× 0.014 =) 0.05 when

Aid/GDP increases by 1.=> leakage rate is 5%

Page 51: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Quantifying the impact of aid on offshore depositsLeakage rate ≡ Aid-induced increase in offshore deposits

Calculating average leakageI βAid = 3.4

I => 1% Aid(%GDP) ↑ => 3.4% haven deposits ↑I mean of haven/GDP: 0.022 (in the high aid sample)I haven/GDP increases by (3.4× 0.022 =) 0.075 when

Aid/GDP increases by 1=> leakage rate is 7.5%

Calculating aggregate leakageI Assume βAid = 3.4 applies uniformly

I mean of haven/GDP weighted by the share of average aidreceived: 0.014

I haven/GDP increases by (3.4× 0.014 =) 0.05 whenAid/GDP increases by 1.=> leakage rate is 5%

Page 52: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Quantifying the impact of aid on offshore depositsLeakage rate ≡ Aid-induced increase in offshore deposits

Calculating average leakageI βAid = 3.4

I => 1% Aid(%GDP) ↑ => 3.4% haven deposits ↑I mean of haven/GDP: 0.022 (in the high aid sample)I haven/GDP increases by (3.4× 0.022 =) 0.075 when

Aid/GDP increases by 1=> leakage rate is 7.5%

Calculating aggregate leakageI Assume βAid = 3.4 applies uniformlyI mean of haven/GDP weighted by the share of average aid

received: 0.014

I haven/GDP increases by (3.4× 0.014 =) 0.05 whenAid/GDP increases by 1.=> leakage rate is 5%

Page 53: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Quantifying the impact of aid on offshore depositsLeakage rate ≡ Aid-induced increase in offshore deposits

Calculating average leakageI βAid = 3.4

I => 1% Aid(%GDP) ↑ => 3.4% haven deposits ↑I mean of haven/GDP: 0.022 (in the high aid sample)I haven/GDP increases by (3.4× 0.022 =) 0.075 when

Aid/GDP increases by 1=> leakage rate is 7.5%

Calculating aggregate leakageI Assume βAid = 3.4 applies uniformlyI mean of haven/GDP weighted by the share of average aid

received: 0.014I haven/GDP increases by (3.4× 0.014 =) 0.05 when

Aid/GDP increases by 1.=> leakage rate is 5%

Page 54: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Leakage increases with aid dependence

0.0

5.1

.15

Impl

ied

leak

age

rate

02

46

810

Poin

t est

imat

e an

d co

nfid

ence

bou

nds

.01 .015 .02 .025 .03Sample threshold (Mean WB aid / GDP)

Model point estimate (left axis) Implied leakage rate (right axis)

I Aggregate leakage in aid-dependent countries: 5%.I Aid might be most susceptible to capture in countries that

need it most

Page 55: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (1/3)

Highly unlikely explanations

I Portfolio adjustments by commercial and centralbanksI data only cover private deposits

I Profit shifting by multinational firmsI data do not cover indirect ownership

I Economic stimulusI GDP growth is already controlled forI output response to stimulus is typically protractedI not clear why money does not flow to non-havens

Page 56: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (1/3)

Highly unlikely explanations

I Portfolio adjustments by commercial and centralbanksI data only cover private deposits

I Profit shifting by multinational firmsI data do not cover indirect ownership

I Economic stimulusI GDP growth is already controlled forI output response to stimulus is typically protractedI not clear why money does not flow to non-havens

Page 57: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (1/3)

Highly unlikely explanations

I Portfolio adjustments by commercial and centralbanksI data only cover private deposits

I Profit shifting by multinational firmsI data do not cover indirect ownership

I Economic stimulusI GDP growth is already controlled forI output response to stimulus is typically protractedI not clear why money does not flow to non-havens

Page 58: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (2/3)

Elite capture

I Consistent withI money flows to havens, but not to non-havensI concurrence aid inflows and capital outflows within the

same quarterI outflows ↑ when corruption ↑ and disclosure requirements ↑

Page 59: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (2/3)

Elite captureI Consistent with

I money flows to havens, but not to non-havens

I concurrence aid inflows and capital outflows within thesame quarter

I outflows ↑ when corruption ↑ and disclosure requirements ↑

Page 60: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (2/3)

Elite captureI Consistent with

I money flows to havens, but not to non-havensI concurrence aid inflows and capital outflows within the

same quarter

I outflows ↑ when corruption ↑ and disclosure requirements ↑

Page 61: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (2/3)

Elite captureI Consistent with

I money flows to havens, but not to non-havensI concurrence aid inflows and capital outflows within the

same quarterI outflows ↑ when corruption ↑ and disclosure requirements ↑

Page 62: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (3/3)

Domestic aid-sponsored contractors depositing money inhavens

I consistent withI larger effects when the financial sector is small

I inconsistent withI larger effects when a greater share of procurement goes to

foreign firmsI money flowing to havens, but not to non-havens

Page 63: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (3/3)

Domestic aid-sponsored contractors depositing money inhavensI consistent with

I larger effects when the financial sector is small

I inconsistent withI larger effects when a greater share of procurement goes to

foreign firmsI money flowing to havens, but not to non-havens

Page 64: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Which mechanisms are at play? (3/3)

Domestic aid-sponsored contractors depositing money inhavensI consistent with

I larger effects when the financial sector is smallI inconsistent with

I larger effects when a greater share of procurement goes toforeign firms

I money flowing to havens, but not to non-havens

Page 65: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Conclusion

I WB aid disbursements coincide with increases inoffshore deposits held in havensI but only in highly aid dependent countries

I Consistent with elite capture

Any feedback you may have would be much appreciated!

Page 66: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Crowding in of aid

[label=crowding in](1) (2)

Sample WB aid >2% WB aid >1%Annual non WB aid (% of GDP) Annual non WB aid (% of GDP)

Annual WB aid (% of GDP) -0.054 -0.128(0.385) (0.261)

Annual GDP (% growth) -0.048* -0.024(0.023) (0.019)

Observations 425 910R-squared 0.608 0.541

Standard errors are clustered by country. Country and time FE included.

Page 67: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Results for Individual Havens

Dependent variable: % Change in haven depositsSwitzerland Luxembourg Belgium Jersey etc

Aid (%GDP) 2.872* 2.717** -1.035 -2.068(1.563) (1.197) (0.726) (2.026)

Observations 2,334 1,796 2,075 1,304R-squared 0.081 0.087 0.083 0.081Standard errors are clustered by country. Country and time FE are included.

Page 68: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Disbursement Profiles

0.0

5.1

.15

.2Pr

edic

ted

Dis

burs

emen

t Sha

re

0 10 20 30 40 50Time (quarters) - (approval quarter=0)

average by sector-region overall average

Disbursement Profiles

IV

Page 69: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Havens

I Based on Andersen et al. (2017)I financial centers identified by the OECD in 2008 as not

providing bank information to foreign governments plusMacao, SAR, China, and Hong Kong, SAR, China

I Bahamas, Bermuda, Cayman Islands, NetherlandsAntilles, Panama, Bahrain, Hong Kong, Macao, Singapore,Austria, Belgium, Guernsey, Isle of Man, Jersey,Liechtenstein, Luxembourg, and Switzerland

Page 70: Elite Capture of Foreign Aid: Evidence from Offshore Bank

Highly aid dependent countries

I Average annual WB aid exceeds 2% of GDP

I Afghanistan, Armenia, Burkina Faso, Burundi, Eritrea,Ethiopia, Ghana, Guinea-Bissau, Guyana, KyrgyzRepublic, Madagascar, Malawi, Mali, Mauritania,Mozambique, Niger, Rwanda, Sao Tome and Principe,Sierra Leone, Tanzania, Uganda, Zambia