Lecture 21: Institutions II�

Dave Donaldson and Esther Duflo

14.73 Challenges of World Poverty

Institutions II: Plan for the lecture

� Discussion of assigned reading (Acemoglu, Johnson and Robinson)

Causes of Long-Run Development

� Authors draw distinction between ‘proximate’ and ‘fundamental’ causes

� Where do the topics we’ve covered so far in this course (food, education, health, technology, finance) fit in to this distinction?

Fundamental Causes

� Authors draw distinction between ‘geography’ and ‘institutions’ as two candidates for fundamental causes of long-run development.

� What could go wrong with drawing this distinction?

� What else might matter for long-run development?

Why Might Geography Matter? 1. Montesquieu: Geography affects humans (and their capacity

for work, thinking and learning) directly.

2. Sachs: Geography affects other living organisms (plants, diseases, disease vectors...) directly.

3. Diamond: Geography affects the types of technologies (e.g. domesticated animals) that can be used, and these may have long-run consequences (e.g. the agricultural revolution).

4. Diamond: Geography affects ability of humans to migrate. � Latitudinally-oriented continents (Eurasia) mean a large

migratory range at similar climates (so can take appropriate technology with you, and will face familiar diseases wherever you go). Longitudinally-oriented continents (eg the Americas) are different.

� Similar argument: a large group of people facing similar environments can take advantage of economies of scale in knowledge-production.

But What are Clear Counter-examples to Geography?

Do We See ‘Geography’ at Work Here?

Do We See ‘Geography’ at Work Here? From Paul Romer’s ‘Charter Cities’ blog

Geography and the Reversal of Fortune

� Geography is (largely) persistent.

� So if geography is a key fundamental determinant of development, development should also be persistent.

� By contrast, institutions change (and changed big time around colonization, c 1600).

Reversal of Fortune I Urbanization as a proxy for GDP today

Figure 3Log income per capita in 1995vs.

urbanization in 1995Lo

g G

DP

per c

apita

, PPP

, 199

5

Urbanization in 19950 50 100

6

8

10

AGO

ARG

AUS

BDI

BEN

BFABGD

BHS

BLZ

BOL

BRA

BRB

BWA

CAF

CAN

CHL

CIVCMR COG

COL

COM

CPV

CRI

DMADOMDZAECU

EGY

ERI

ETH

FJI

GAB

GHAGIN

GMB

GRDGTM

GUY

HKG

HND

HTI

IDN

IND

JAM

KEN

KNA

LAO

LCA

LKA

LSO

MAR

MDG

MEX

MLIMOZ

MRT

MUS

MWI

MYS

NAM

NER NGA

NIC

NPL

NZL

PAK

PAN

PER

PHLPRY

RWA

SDN SEN

SGP

SLE

SLVSURSWZ

TCD

TGO

TTO

TUN

TZA

UGA

URY

USA

VCT

VEN

VNM

ZAF

ZAR ZMB

ZWE

Reversal of Fortune II Only ex-colonies plotted

Log

GD

P pe

r cap

ita, P

PP, 1

995

Urbanization in 15000 5 10 15 20

7

8

9

10

ARG

AUS

BGD

BLZ

BOL

BRA

CAN

CHL

COL CRI

DOM DZAECU

EGY

GTM

GUY

HKG

HND

HTI

IDN

IND

JAM

LAO

LKA

MAR

MEXMYS

NIC

NZL

PAK

PAN

PER

PHLPRY

SGP

SLV

TUN

URY

USA

VEN

VNM

Figure 4Log income per capita in 1995 vs. urbanization

in 1500

Reversal of Fortune III Only ex-colonies plotted

Figure 5Log income per capita in 1995 vs. log

population density in 1500Lo

g G

DP

per c

apita

, PPP

, 199

5

Log Population Density in 1500-5 0 5

6

8

10

AGO

ARG

AUS

BDI

BEN

BFA BGD

BHS

BLZ

BOL

BRA

BRB

BWA

CAF

CAN

CHL

CIVCMRCOG

COL

COM

CPV

CRI

DMADOM DZAECU

EGY

ERI

ETH

GAB

GHAGIN

GMB

GRDGTM

GUY

HKG

HND

HTI

IDN

IND

JAM

KEN

KNA

LAO

LCA

LKA

LSO

MAR

MDG

MEX

MLIMOZ

MRT

MWI

MYS

NAM

NER NGA

NIC

NPL

NZL

PAK

PAN

PER

PHLPRY

RWA

SDNSEN

SGP

SLE

SLVSUR SWZ

TCD

TGO

TTO

TUN

TZA

UGA

USA

VCT

VEN

VNM

ZAF

ZARZMB

ZWE

Two Important Qustions

1. Is the ‘reversal of fortune’ necessarily evidence against the geography hypothesis?

2. Is the ‘reversal of fortune’ necessarily evidence in favor of the institutions hypothesis?

Reversal of Fortune—Timing Industrial revolution: industry is particularly ‘institution-sensitive’ (huge up-front costs of innovating and investing) Figure 6

Timing of the ReversalUrbanization in excolonies with low and high urbanization in 1500

(averages weighted within each group by population in 1500)

0

5

10

15

20

25

800 1000 1200 1300 1400 1500 1600 1700 1750 1800 1850 1900 1920

low urbanization in 1500 excolonies high urbanization in 1500 excolonies

The Industrial Revolution Industrial revolution is key to long-run development Figure 7

Reversal, Industrialization and DivergenceIndustrial Production Per Capita, UK in 1900 = 100

(from Bairoch)

0

50

100

150

200

250

300

350

400

1750 1800 1830 1860 1880 1900 1913 1928 1953

US Australia Canada New Zealand Brazil Mexico India

Colonialization and the Reversal of Fortune I

� Colonialization was process that changed institutions a great deal.

� Who created the new institutions? The colonizers.

� What institutions did they create? (Good or bad?)

Colonialization and the Reversal of Fortune II

� The institutions the colonizers created depended on what was in their own self-interest.

� Self-interest depended on whether settlers present or not. � If settlers: A bit like quasi-democracies back home—need to

keep the median ‘voter’ happy. � If no settlers: Attempt to extract as much as possible from the

colony. Exploit native populations. Coerce labor into slavery.

� So what determined whether there were settlers or not? � Could settlers survive there? Disease, etc. � Were settlers needed there? Some resources just needed loads

of unskilled labor, and only a few settlers to manage the process.

Settler Mortality

� Thinking in this manner generates another ‘experiment’ in long-run development:

� Suppose that whether settlers were killed by the disease environment when they arrived or not (we call this: ‘settler mortality’) was basically random.

� Then places with high settler mortality should have received fewer settlers, and hence received bad institutions.

� Then, if (for whatever reason) institutions are persistent (bad institutions beget bad institutions), then these places will have bad institutions today.

� Then, if bad institutions (PR) are bad for development (Y ), places with high settler mortality (SM) should have low levels of development today.

� Acemoglu, Johnson and Robinson (2001) pursue this logic empirically.

Settler Mortality as an Instrumental Variable

� We are interested in the question: Does PR cause Y ?

� The correlation between PR and Y is strongly positive. But we are worried that this correlation does not prove causation.

� Suppose that SM affects PR through the incentives for colonizers to build good institutions (and the fact that institutions are persistent).

� Suppose further that SM affects Y only because SM affects PR and PR affects Y .

� Then SM can be used as an instrumental variable for PR in the regression of Y on PR.

The Question: Does PR affect Y ?

Figure 1Log income per capita in 1995 vs. perceived

protection against expropriation risk, 1985-95.

Log

GD

P pe

r cap

ita in

199

5

Average Protection Against Expropriation Risk, 1985-19952 4 6 8 10

6

8

10

.

.

AGO

ARE

ARG

AUS AUTBEL

BFA BGD

BGR

BHR

BHS

BOL

BRABWA

CANCHE

CHL

CHN

CIVCMRCOG

COLCRI

CZE

DNK

DOM DZAECU

EGY

ESP

ETH

FINFRA

GAB

GBR

GHAGIN

GMB

GRC

GTM

GUY

HKG

HND

HTI

HUN

IDN

IND

IRL

IRN

ISLISR

ITA

JAMJOR

JPN

KEN

KOR

KWT

LKA

LUX

MAR

MDG

MEX

MLI

MLT

MNG

MOZMWI

MYS

NER NGA

NIC

NLDNOR

NZL

OMN

PAK

PAN

PER

PHL

POL

PRT

PRY

QAT

ROM RUS

SAU

SDNSEN

SGP

SLE

SLVSUR

SWE

SYR

TGO

THATTO

TUNTUR

TZA

UGA

URY

USA

VEN

VNM

YEM

ZAF

ZAR ZMB

ZWE

The Instrument (First Stage): SM affects PR

Figure 8 Perceived protection against expropriation

risk, 1985-95 vs. log settler mortality.

Aver

age

Prot

ectio

n Ag

ains

t Exp

ropr

iatio

n, 1

985-

95

Log Settler Mortality2 4 6 8

4

6

8

10

AGO

ARG

AUS

BFA

BGD

BHS

BOL

BRA

CAN

CHL

CIV

CMR

COG

COLCRI

DOM

DZAECUEGY

ETH

GAB

GHAGIN

GMB

GNB

GTM

GUY

HKG

HND

HTI

IDN

IND

JAM

KENLKA

MAR

MDG

MEX

MLI

MMR

MYS

NER

NGANIC

NZL

PAKPAN

PER

PNGPRY

SDN

SEN

SGP

SLE

SLV

SUR

TGO

TTO

TUNTZA

UGA

URY

USA

VEN

VNM

ZAF

ZAR

The Instrument (Reduced Form): SM affects Y If SM affects Y only because SM affects PR, then SM is a valid IV. Further, the ratio of the reduced form slope to the first stage slope is the causal effect of PR on Y .

Figure 9Log income per capita in 1995 vs. log settler

mortality.

Log

GD

P pe

r cap

ita in

199

5

Log Settler Mortality2 4 6 8

6

8

10

AGO

ARG

AUS

BDI

BEN

BFABGD

BHS

BLZ

BOL

BRA

BRB

CAF

CAN

CHL

CIVCMRCOG

COLCRI

DOMDZAECU

EGY

ETH

FJI

GAB

GHAGIN

GMB

GTM

GUY

HKG

HND

HTI

IDN

IND

JAM

KENLAO

LKA

MAR

MDG

MEX

MLI

MRT

MUSMYS

NER NGA

NIC

NZL

PAK

PAN

PERPRY

RWA

SDNSEN

SGP

SLE

SLVSUR

TCD

TGO

TTO

TUN

TZA

UGA

URY

USA

VEN

VNM

ZAF

ZAR

What Could Violate this Logic?

Application to Development Policy Today?

� How useful is the ‘geography vs institutions’ debate for contemporary development policy?

� How useful is the AJR empirical approach for contemporary development policy?

� Go back in time and save more settlers? � Right the wrongs of past colonial injustice?

� If institutions matter, what can be done to make them better?

MIT OpenCourseWarehttp://ocw.mit.edu

14.73 The Challenge of World Poverty Fall 2009 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.