3 9080 01917 6038
ml
m.
i^zIiifiiiiftiVf; :•.
ililiil
«
-iMiii
iH)
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium IVIember Libraries
http://www.archive.org/details/colonialoriginsoOOacem
DEWEY
iS)
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
THE COLONIAL ORIGINS OF COMPARATIVE
DEVELOPMENT: AN EMPIRICAL INVESTIGATION
Daron Acemoglu, MIT
Simon Johnson, MIT
James A. Robinson, UCal, Berkeley
Working Paper 00-22
September 2000
Room
E52-251
50 Memorial Drive
Cambridge, MA 02142
downloaded without charge from the
Science Research Network Paper Collection at
This paper can be
Social
http:/ /papers. ssrn.com/paper.taf?abstract_id=244582
roCT
3
2000
LIBRARIES
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
THE COLONIAL ORIGINS OF COMPARATIVE
DEVELOPMENT: AN EMPIRICAL INVESTIGATION
Daron Acemoglu, MIT
Simon Johnson, MIT
James A. Robinson, UCal, Berkeley
Working Paper 00-22
September 2000
Room E52-251
50 Memorial Drive
Cambridge, MA 02142
This paper can be downloaded without charge from the
Network Paper Collection at
http: //papers, ssrn. com/paper. taf?abstract_id=244582
Social Science Research
The
Colonial Origins of Comparative
Development:
An
Daron Acemoglu^
Simon Johnson^
Empirical Investigation^
July
7,
James A. Robinson^
2000
Abstract
We
European colonialists to
on economic performance. Our argument is that
exploit differences in the mortality rates faced by
estimate the effect of institutions
Europeans adopted very
different colonization policies in different colonies,
different associated institutions.
The
with
choice of colonization strategy was, at least in
determined by whether Europeans could settle in the colony. In places where
Europeans faced high mortality rates, they could not settle and they were more likely
to set up worse (extractive) institutions. These early institutions persisted to the
present. We document evidence supporting these hypotheses. Exploiting differences
in mortality rates faced by soldiers, bishops and sailors in the colonies in the 17th,
18th and 19th centuries as an instrument for current institutions, we estimate large
effects of institutions on income per capita. Our estimates imply that differences in
institutions explain approximately three-quarters of the income per capita differences
part,
across former colonies.
Once we
control for the effect of institutions,
we
find that
countries in Africa or those closer to the equator do not have lower incomes.
*We thank
John Gallup, Chad Jones, Richard Locke.
and participants at the Harvard-MIT Development Economics seminar, Berkeley Political
Science seminar, Canadian Institute for Advanced Research, Columbia Social Sciences seminar, the Stanford Social Science History Institute conference, and the Toulose Political Economy conference for useful
comments. We also thank Robert McCaa for guiding us to the data on bishops' mortality.
t Massachusetts Institute of Technology, Department of Economics, E52-371, Cambridge, MA 02319,
and Canadian Institute for Advanced Research; e-mail: daron@mit.edu
f Massachusetts Institute of Technology, Sloan School of Management, Cambridge, MA 02319; e-mail:
s j ohnsonSmit edu.
^University of California, Department of Political Science, 210 Barrows Hall. Berkeley CA94720; e-mail:
jamesarOsocrates.berkeley .edu.
Andrei
.Joshua Angrist, Abhijit Baneijee, Esther Duflo,
Shleifer,
.
1
Introduction
1
What
are the fundamental causes of the large differences in income per capita across
countries? Although there
in institutions
is still little
and property
consensus on the answer to this question, differences
rights have received considerable attention in recent years.
Countries with better "institutions", more secure property rights, and
more
policies will invest
in physical
and human
efficiently to achieve a greater level of
income
and
capital,
less distortionary
will use these factors
North and Thomas, 1973, North,
(e.g..
This view receives some support from cross-country correlations
1981, Jones, 1981).
between measures of property rights and economic development
1995, Mauro, 1995, Hall
Knack and
(e.g..
Keefer,
and Jones, 1999, Rodrik, 1999), and from a few micro-studies
that investigate the relationship between property rights and investment or output
Besley, 1995, Mazingo, 1999, Johnson,
At some
more
level, it is
(e.g.,
McMillan and Woodruff, 1999).
obvious that institutions matter. Witness, for example, the divergent
paths of North and South Korea, or East and West Germany, where one part of the country
stagnated under central planning and collective ownership, while the other prospered with
we
private property and a market economy. Nevertheless,
still
lack conclusive evidence
that institutional differences can have a large enough effect to explain the phenomenal
differences in output per capita across countries.
rich choose or
quite likely that economies that are
Perhaps more important, economies that are
can afford better institutions.
different for a variety of reasons will differ
It is
both
in their institutions
and
in their
income
per capita.
To estimate the impact
exogenous variation
differences
among
of institutions on economic performance,
in institutions.
we propose a theory
In this paper,
of
of institutional
countries colonized by Europeans, and exploit this theory to derive a
possible source of exogenous variation.
1.
we need a source
Our theory
rests
on three premises:
There were
different types of colonization policies
institutions.
At one extreme, European powers
set
up
which created different sets of
"extractive states", exemplified
by the Belgian colonization of the Congo. These institutions did not introduce much
protection for private property, nor did they provide checks and balances against
government expropriation.
transfer as
much
In fact, the
main purpose
of the extractive state
was
to
of the resources of the colonv to the colonizer.^
^Young (1994) and Chazan
et al.
(1993) argue that the colonization policies of
European powers had
At the other extreme, many Europeans went and
creating
what the historian Alfred Crosby (1986)
tried to replicate
European
settled in a
calls
institutions, with great
"Neo-Europes"
2.
The
.
of colonies,
The
settlers
emphasis on private property,
and checks against government power. Primary examples of
New
number
this include Australia,
Zealand, Canada, and the United States.
colonization strategy was influenced by the feasibility of settlements. In partic-
ular, in places
where the disease environment was not favorable to European
settle-
ment, the cards were stacked against the creation of Neo-Europes, and the formation
3.
of the extractive state
was more
The
institutions persisted even after independence.
colonial state
and
likely.
Based on these three premises, we use the mortality rates expected by the
pean
settlers in the colonies as
More
specifically,
an instrument
Euro-
for current institutions in these countries.^
our theory can be schematically summarized as
(potential) settler
early
,
=> settlements
=^
=r'
mortality
We
first
institutions
current
=^
current
performance
institutions
use data on the mortality rates of soldiers, bishops, and sailors stationed in the
colonies between the 17th and 19th centuries, largely based on the
work of the historian
Philip Curtin. These give a good indication of the mortality rates faced by settlers. Eu-
ropeans were well informed about these mortality rates at the time, though they did not
know how
to control the diseases that caused these high mortality rates.
since these mortality rates refer to fairly
countries.
We
homogeneous groups, they
document empirically that
Furthermore,
are comparable across
(potential) settler mortality rates were a
major
determinant of settlements; that settlements were a major determinant of early institutions (in practice, institutions in 1900); that there
institutions
effect
and institutions today; and
is
a strong correlation between early
finally that current institutions
have a first-order
on current performance.'^
a long-lasting effect on Africa. In contrast, Chabal (1986) and Herbst (2000) maintain that the current
.African state is a continuation of the precolonial state. Although our assessment agrees with that of
Young, since we are not comparing ex-colonies to non-colonized countries, our empirical approach does
not take a position on this.
"Note that although only some countries were colonized, there is no selection bias here. This is because
the question we are interested in is the effect of colanization policy conditional on being colonized.
^We do not argue that differences in mortality rates are the only, or even the main, cause of variation
in institutions.
For our empirical approach to work, all we need is that they are a source of exogenous
variation.
Our most parsimonious specification
tutions,
and instrument the
rights cind checks against
latter
by
to regress current performance on current insti-
is
our focus
settler mortality rates. Since
government po^er. we use the protection against
priation" index from Pohtical Risk Services as a proxj- for institutions.
differences in institutions originating
first-stage relationship
from
different types of states
between this measure of institutions and
is
on
property-
Trisk of expro-
This measures
and state
policies.
The
settler mortahty" is strong.
For example, settler mortality alone explains over 25 percent of the ^^ariation in this index
of institutions. Using this specification,
of per capita income.
differences
We
find that iostitutions are a
New
this relationship is
differences in institutions.
not driven by
outliers.
rest of the
For example, excluding
Zealand. Canada, and the United States does not change the results,
economic performance
dummy
OLS
that approximately three-quarters of the cross-countty income
nor does excluding Africa. Interestingly, we show that once the
the
major determinant
quite precise, and in fact larger than the
we observe can be explained by
document that
Australia,
The estimates are
They suggest
estimates.
we
for Africa
is
is
effect of institutions
on
controlled for, neither distance from the equator (latitude) nor
significant.
These
results suggest that Africa is poorer
than the
world not because of pure geographic or cultural factors, but because of worse
institutions.
The
validity- of
our approach
is
threatened
if
other factors correlated with the estimates
of the mortahty rates faced by the settlers affect income per capita.
We adopt two strategies
to substantiate that our results are not driven by omitted factors.
whether institutions have an important
effect
potentially correlated Tvith settler mortality
and economic outcomes.
controls for climate, geography, rehgion, legcd origin,
Furthermore, the results are
soil quality.
we
investigate
once we control for a number of
of these overturn our results; the estimates change remarkably
and
First,
eilso
mcdn
We
little
\-ariables
find that
when we
none
include
colonizer, natural resources,
robust to the inclusion of controls
for the current disease en\ironment (e.g.. the pre^-alence of malaria), the current fraction
of the population of European descent, and measures of ethnoliaguistic fragmentation.
Interestingly,
we
find that once
we
control for the effect of settler mortality on institutions,
the identity of the colonial power has no effect on institutions or on income per capita.
Naturally,
it is
impossible to control for
all
possible ^ariables that might be correlated
with settler mortality- and economic outcomes. Furthermore, our empirical approach might
capture the
factors.
effect of settler mortality
on economic performance, but working through other
For example, early European settlers might have brought a
""culture"
conducive
to economic progress, which could
still
have an
effect
on income per capita today.
Our
instrumental variables strategy would then incorrectly assign this effect to institutions.
We
is
deal with this problem by using a simple overidentification test.
Since our hypothesis
that settler mortality affected settlements; settlements affected early institutions; and
we can
early institutions persisted and formed the basis of current day institutions,
validity of our approach by using measures of
early institutions as additional instruments.
whether
settler mortality, or
performance.
The
rates
institutions,
effect
on current
no evidence
on economic outcomes.
who have
pointed out the link between settler mortality
though scholars such as McNeill (1976), Crosby (1986) and Diamond
(1997) have discussed the influence of diseases on
ticular,
and of
then use overidentification tests to detect
results are encouraging for our approach; they generate
are not aware of others
and
We
to the colonies
any of the other instruments, has a direct
for a direct effect of settler mortality
We
European migration
test the
human
history.
emphasizes comparative development, but his theory
is
Diamond
(1997), in par-
based on the geographical
determinants of the incidence of the neolithic revolution. He ignores both the importance
more recent development, which
of institutions and the potential causes of divergence in
are the
main focus
of our paper.
Work by Gann and Duignan
(1962), Robinson
and
Gallagher (1961), Denoon (1983), and Cain and Hopkins (1993) emphasizes that settler
colonies such as the U.S. and
New
Zealand are different from other colonies, and point
out that these differences were important for their economic success.
literature does not develop the link
Our argument
is
most
between mortality, settlements and institutions.
closely related to
work on the influence of colonial experience
on institutions. Hayek (1960) argued that the British
to the French civil law, which
common
law tradition was superior
was developed during the Napoleonic era to
More
interference with state policies (see also Lipset, 1994).
de-Silanes, Shleifer
Nevertheless, this
recently.
restrain judges'
La Porta, Lopez-
and Vishny (1998, 1999) emphasize and document the importance of
colonial origin (the identity of the colonizer)
example, they show that
common
and
legal origin
on current
institutions.
For
law countries and former British colonies have better
property rights and more developed financial markets.
Similarly, North,
Weingast (1998) argue that former British colonies prospered
Summerhill and
relative to former French,
Spanish and Portuguese colonies because of the good economic and political institutions
they inherited from Britain, and Landes (1998, chapters 19 and 20) stresses the importance
of the culture inherited from Britain in these colonies.
which focuses on the identity of the colonizer, we emphasize
In contrast to this approach,
the conditions in the colonies.
Specifically, in our theory
where they could not
our argument
settle,
related to that of
is
institutions, but link
data
in the
—
it is
not the identity of the colonizer or legal
whether European colonialists could
origin that matters, but
location:
— and
them
Empirically, our work
to factor
is
safely settle in a particular
they brought worse institutions.
Engerman and
Sokoloff (1997)
In this respect,
who
emphasize
also
endowments.
related to a
number
of other attempts to uncover the link be-
tween institutions and development, as well as to Bertocchi and Canova (1996) and Grier
(1999)
who
investigate the effect of being a colony on postwar growth.
Two
papers deal
with the endogeneity of institutions by using an instrumental variables approach as we do
Mauro
here.
and
(1995) instruments for corruption using ethnolinguistic fragmentation. Hall
-Jones (1999), in turn, use latitude (distance
from the equator) as an instrument
social infrastructure because, they argue, latitude
is
correlated with 'Western influence',
which leads to good institutions. The theoretical reasoning
entirely convincing.
Western influence
It is
in the
for
for these instruments
is
not
not easy to argue that the Belgian influence in the Congo, or
Gold Coast during the era of slavery promoted good institutions
or governance. Ethnolinguistic fragmentation, on the other hand, seems endogenous, especially since such fragmentation almost completely disappeared in
of growth
when a
centralized state
1983). Econometrically, the
sibly have a direct
eff"ect
and market emerged
problem with both studies
(see, e.g.,
is
Europe during the era
Weber, 1976, Anderson,
that their instruments can plau-
on performance. For example. Easterly and Levine (1997) argue
that ethnolinguistic fragmentation can affect performance by creating political instability,
while
Bloom and Sachs
effect of climate
pedigree;
it
(1998) and Gallup, Mellinger, and Sachs (1998) argue for a direct
on performance. In
climate theory of development has a long
goes back at least to Montesquieu [1748](1989),
and despotism are more
effect,
fact, this
likely in
warmer
climates.
If,
who
indeed, these variables have a direct
they are invalid instruments and do not establish that
The advantage
of our approach
settler-mortality
than through
is
it is
institutions that matter.
that conditional on the variables
more than 100 years ago should have no
its effect
suggested that low income
effect
we already
control for,
on output today, other
on institutions. Interestingly, our results show that distance from
the equator does not have an independent effect on economic performance, so
it
validates
the use of this variable as an instrument in the work by Hall and Jones (1999).
The next
section outlines our hypothesis
Section 3 presents
OLS
regressions of
and provides supporting
historical evidence.
GDP per capita on our index of institutions.
Section 4
describes our key instrument for institutions, the mortality rates faced by potential settlers
at the time of colonization. Section 5 presents our
main
results. Section 6 investigates the
robustness of our results, and Section 7 concludes.
The Hypothesis and Historical Background
2
We
hypothesize that settler mortality affected settlements; settlements affected early
stitutions;
and early
this section,
link
we
institutions persisted
and formed the basis of current
discuss and substantiate this hypothesis.
in-
In
institutions.
The next subsection
discusses the
between mortality rates of settlers and settlement decisions, then we discuss differences
in colonization policies,
and
finally,
we turn
to the causes of institutional persistence.
Mortality and settlements
2.1
There
is little
doubt that mortality rates were a key determinant of European settlements.
Curtin (1964 and 1998) documents
how both
public of the mortality rates in the colonies.
settle in
West Africa foundered due
Freedom" European mortality
Company
For example, early European attempts to
to high mortality from disease.
in the first
April 1793) there was 61 percent mortality
Sierra Leone
the British and French press informed the
year was 46 percent, in
among Europeans, and
In the "Province of
Bulama
(April 1792-
in the first year of the
(1792-1793) 72 percent of the European settlers died.
On Mungo
Park's Second Expedition (May-November 1805), 87 percent of Europeans died during the
overland trip from
Gambia
to the Niger,
and
all
the Europeans died before completing the
expedition. Such rates of mortality were shockingly high for Europeans at the time.''
An
interesting
Pilgrim fathers.
example of the awareness of the disease environment comes from the
They decided
very high mortality rates in
to migrate to the U.S. rather than
Guyana
(see Crosby, 1986, pp.
comes from the Beauchamp Committee
British convicts,
who had
in
Guyana because
143-144).
of the
Another example
1795 which was set up to decide where to send
One
previously been sent to the U.S..
was the island of Lemane, 400 miles up Gambia
The committee
river.
''Most mortality in the tropics was from (1) malaria (particularly
yellow fever (with devastating periodic epidemics).
In the
first
of the leading proposals
rejected this
Plasmodium Falciporum), and
(2)
half of the nineteenth century there
—
was almost a complete misunderstanding of the nature of malaria
"miasma" from swamps was the
prevailing view.
Quinine was available but not understood nor used widely. The role of hygiene was
also not properly understood.
In the second half of the nineteenth century, there developed improved
heuristic rules about
less at
how
higher altitudes.
to control disease
Low
—
e.g.,
an understanding that mortality from malaria
is
Throughout the whole nineteenth century, areas without malaria, such as
Zealand or Mauritius, were more healthy than Europe.
early twentieth century.
often
mortality for Europeans in areas with tropical diseases only arrived in the
New
possibility precisely because they decided mortality rates
convicts.
would be too high even
South- West Africa was also rejected for health reasons. The
for the
final decision
was
to send convicts to Australia.
The eventual expansion
In places
there.
many
where the early
new
incentive for
of
settlers to
of the colonies
settlers,
less
Curtin (1964), for example, documents how early
British expectations for settlement in
among attempted
also related to the living conditions
high mortality rates, there would be
settlers faced
come.
was
West Africa were dashed by very high mortality
about half of
whom
could be expected to die in the
first
year.
Types of colonization and settlements
2.2
The
historical evidence supports
both the notion that there was a wide range of
European
ent types of colonization and that the presence or absence of
key determinant of the form colonialism took.
(1962),
Historians, including
settlers
differ-
was a
Gann and Duignan
Robinson and Gallagher (1961), Denoon (1983), and Cain and Hopkins (1993),
have documented the development of "settler colonies" where Europeans settled
,
numbers, and
was modeled
life
settler colonies
after the
home
country.
Denoon
(1983) emphasizes that
had representative institutions which promoted what the
and what they wanted was freedom and the
ability to get rich
in large
settlers
wanted,
by engaging in trade. He
argues that "there was undeniably something capitalist in the structure of these colonies.
Private ownership of land and livestock was well established very early..."
In
many
cases,
when
the establishment of European-like institutions did not arise
naturally, the settlers were ready to fight for
try.
Australia
is
them against the wishes
an interesting example here.
Most
were ex-convicts, but the land was owned largely by
were therefore unequal, and there
of landowners.
v/as
first,
home
coun-
of the early settlers in Australia
ex-jailors.
Initial
property rights
But, the majority of settlers wanted institutions and political rights
and
electoral representation.
They demanded jury
trials,
(see
like
freedom from arbi-
Although the British government
the settlers argued that they were British and deserved the
home country
of the
no legal protection against the arbitrary power
those prevailing in England at the time.
trary arrest,
35).^
(p.
same
resisted at
rights as in the
Hughes, 1987). Cain and Hopkins agree with this conclusion and write
^Bates (1983, ch. 3) gives a nice example of the influence of settlers on policy. The British colonial
government pursued many policies that depressed the price of cocoa, the main produce of the farmers in
Ghana. In contrast, the British government supported the prices faced by the commercial cereal farmers
in Kenya. Bates shows that this was mainly because in Kenya, but not in Ghana, there were many settler
farmers, who exerted considerable pressure on policy.
—
bowed
(1993, p. 237) "from the late 1840's the British
to local pressures and, in line with
observed constitutional changes taking place in Britain herself, accepted the idea that, in
mature
colonies, governors should in future
elected legislatures."
after
1870
[in
New
They
form ministries from the majority elements
enormous boom
also suggest that "the
Zealand]...
was an attempt to build up an
in public
in
investment
infrastructure... to
maintain
high living standards in a country where voters expected politicians actively to promote
their
economic welfare."
This
is
in
(p.
225).
sharp contrast to the colonial experience
and 18th centuries, and
main objective
in
in
Latin America during the 17th
Asia and Africa during the 19th and early 20th centuries. The
of the Spanish and the Portuguese colonization
was to obtain gold and other
valuables from America. For example, soon after the conquest the Spanish crown granted
rights to land
and labor
(the encomienda)
and
set
up a complex mercantilist system
monopolies and trade regulations to extract resources from the colonies
(see
of
Lockhart and
Schwartz, 1983, and Lang, 1975).^ Other European powers were attracted to colonialism
because of the success of this strategy (see for example. Young, 1994,
European powers developed the
slave trade in Africa for the
p.
same
64).
reasons.
Before
the mid-nineteenth century, colonial powers were mostly restricted to the African coast
and concentrated on monopolizing trade
witness the
names used
to descibe
in slaves, gold
West African
and other valuable commodities
countries: the
Gold Coast, the Ivory Coast.
Thereafter, colonial policy was driven in part by an element of superpower rivalry, but
mostly by economic motives. Crowder (1968,
that Britain's largest colony on the
50), for
p.
West Coast
example, notes
[Nigeria] should
"it is
significant
have been the one where
her traders were most active and bears out the contention that, for Britain. ...flag followed
trade."
^
Davis and Huttenback (1986,
p.
307) conclude that "the colonial Empire provides
strong evidence for the belief that government was attuned to the interests of business
and willing to divert resources to ends that the business community would have found
profitable."
They
find that before 1885 investment in the British
empire had a return 25
percent higher than that on domestic investment, though afterwards the two converged.
^Migration to Spanish America was limited by the Spanish Crown, in part because of a desire to keep
example Coatsworth, 1982). This gives further
control of the colonists and limit their independence (for
support to our notion that settlers were able to influence the type of institutions set up
even against the wishes of the home country government.
'^Although in almost
and obtain
all
cases the
main objective of
colonial policies
profits, the recipients of these profits varied.
Belgian case,
it
was King Leopold, and
in the colonies,
was to protect economic interests,
it was the state, in the
In the Portuguese case,
in the British case,
it
was often private enterprises who obtained
concessions or monopoly trading rights in Africa (Crowder, 1968, Part
III).
—
Roberts (1976,
193) summarizes the extent of resource extraction by Britain from
p.
Northern Rhodesia by writing
pounds
in taxes
"[from]..
1930 to 1940 Britain had kept for
2,400,000
from the Copperbelt, while Northern Rhodesia received from Britain
only 136,000 pounds in grants for development." Patrick
between 1905 and 1914, 50 percent of
Young
itself
GDP
in
Manning
Dahomey was
(1982) estimates that
extracted by the French, and
(1994, p. 125) notes that taxation rates in Tunisia were four times as high as those
in metropolitan France.
Probably the most extreme case of extraction was that of King Leopold of Belgium
Gann and Duignan
the Congo.
(1979, p. 30) argue that following the
in
example of the Dutch
Leopold's philosophy was that "the colonies should be exploited, not by the
in Indonesia,
operation of a market economy, but by state intervention and compulsory cultivation of
cash crops to be sold to and distributed by the state at controlled prices." Peemans (1975)
documents the amount of resources extracted from the Belgian Congo and calculates that
tax rates on Africans approached 60 percent of their income during the 1920's and 1930's.
Jewsiewicki (1983) writes that during the period
when Leopold was
directly in charge,
human
resources," with a
policy was "based on the violent exploitation of natural and
consequent "destruction of economic and social
life...
dismemberment
[and].,
of political
structures."
In non-settler colonies, there were also few constraints on state power.
powers set-up authoritarian and absolutist
states,
control and facilitating the extraction of resources.
official in
Africa as "the European
mission... to
won out
up
is
summarizes
this as:
"In
colonial
with the purpose of solidifying their
Young
(1994, p. 101) quotes a French
not posted to observe nature,...
impose regulations, to limit individual
84)
(1988, p.
commandant
The
liberties....,
to collect taxes."
He has a
Manning
Europe the theories of representative democracy
But
over the theorists of absolutism...
in Africa, the
European conquerors
set
absolutist governments, based on reasoning similar to that of Louis XIV."
With a
strategy of exploitation in mind, European powers had
in institutions or in infrastructure in Africa.
of return, almost
1994).
(e.g.,
little
incentive to invest
In fact, despite apparently very high rates
no investment went to Africa (except from South Africa, see Freiden,
The Indian
textile industry
Fieldhouse, 1999).
Young
was similarly rundown
writes
"
heavily, with
[the Belgian companies]
a mere 8000 pounds... [to the Congo basin]
— and instituted
provoke an embarrassing public-protest campaign
in
brought
no investment
little
capital
a reign of terror sufficient to
Britain and the United States at a
time when the threshold of toleration for colonial brutality was high." (1994,
p.
104).
Institutional persistence
2.3
There
is
a variety of historical evidence, as well as our regressions in Table 3 below,
suggesting that the control structures set up in the non-settler colonies during the colonial
era persisted, while there
is little
doubt that the institutions of law and order and private
New
property established during the early phases of colonialism in Australia, Canada,
Zealand, and the U.S. have been the basis of the current day institutions of these countries.*
Young emphasizes
that the institutions set up by the colonialists persisted long after the
colonial regime ended.
He
writes that (1994, p.
283) "although
we commonly described
the independent polities as 'new states', in reality they were successors to the colonial
regime, inheriting
its
structures, its quotidian routines
and
practices,
and
its
more hidden
normative theories of governance." Arthur Lewis gives a succinct statement of the issues
(1965, pp. 32-33)
"...for
most of
[the
governments of newly independent African
states] inde-
pendence means merely that they have succeeded to the autocracy vacated by
British and French civil servants.
They model themselves on the arrogant and
arbitrary pattern set by Governors and district commissioners."
There are a number of intuitive economic mechanisms that
persistence of this type.
1.
Here,
we
will lead to institutional
discuss three possibilities.
Setting up functioning institutions, which place restrictions on government power
and respect property
rights, is costly (see, e.g.,
Acemoglu and
these costs have been sunk by the colonial powers, then
independence to switch from this
contrast,
when
the
new
elites inherit extractive institutions,
the existing extractive institutions for their
interesting case
ritius.
^The
not pay the
own
may
they
may
settled in relatively large
thesis that institutions persist for a long
In
not want to
instead prefer to exploit
benefits.
where functioning institutions proved to be very durable
The French
If
elites at
set of institutions to extractive institutions.
incur the costs of introducing better institutions, and
An
may
it
Verdier, 1998).
numbers
in Mauritius,
is
Mau-
and the
in-
time goes back at least to Wittfogel (1957), who argued
that the control structures set up by the large "hydraulic" empires such as China, Russia, and the
Ottoman
Empire persisted for more than 500 years to the 20th century. Engerman and Sokoloff (1997), North,
Summerhill and Weingast (1998), Coatsworth (1999) and La Porta, Lopez-de-Silanes, Shleifer and Vishny
(1998, 1999) also argue that colonial institutions persisted. Engerman, Mariscal and Sokoloff (1998)
provide further evidence supporting this view.
10
stitutions were less extractive than in
many
other colonies.^
After independence,
Paul Berenger and his party Mouvement Militant Mauricien, which were viewed as
Communists
at the time,
came
But
to power.
in contrast to other African regimes,
they continued to support property rights and businesses. In
expanded the export processing zones, which were instrumental
growth experience of Mauritius
2.
The
gains to an extractive strategy
this elite
may
set
(see, for
is
the size of the ruling
elite.
When
a larger share of the revenues, so the elite
In
many
cases where
European powers
institutions, they delegated the day-to-day running of the state
to a small domestic elite. This
after
may depend on
have a greater incentive to be extractive.
up authoritarian
for the very rapid
example. Bowman, 1991).
member would have
small, each
they significantly
fact,
narrow group often was the one to control the state
independence and favored extractive institutions.
Reno
(1995), for example,
argues that the governments of post-independence Sierra Leone adopted the tactics
and institutions of the British colonizers to cement
resources from the rest of society.
modern
evolution of the
is
power and extract
their political
Boone (1992) provides a
state in Senegal. Perhaps the
similar analysis of the
most extensively studied case
the Congo. Most scholars view the roots of authoritarianism under
colonial state practices (e.g., Callaghy, 1984, or.
p.
Mobutu
in the
Turner and Young, 1985, especially
43)
The
situation in Latin
countries
came
America
is
similar.
Independence of most Latin American
in the early nineteenth centuries as
domestic
elites
took advantage of
the invasion of Spain by Napoleon to capture the control of the state. But, the only
thing that changed was the identity of the recipients of the rents
(see, for
example,
Coatsworth, 1978, or Lynch, 1986).^°
3.
Finally,
if
agents
make
irreversible investments that are
ticular set of institutions, they will then
®
Unfortunately, Mauritius
measure of current
relatively rich country, with
measured
in
is
institutions.
complementary to a par-
be more willing to support them, making
not in our base sample as
it
is
not included in Political Risk Services'
Mauritius had very low mortality according to our sources, and
GDP
per capita similar to that of Argentina.
It
is
a
has good institutions
terms of constraint on the executive and rule of law (alternative measures for which we
report results in Appendix Table 4a and 4d.)
came through a virtual coup d'etat by the
Mexico from the hberalizing process under way in
the Mother country... The principal proponent of these conservative efforts was a limited social group of
major landowners and industrialists in the center of the country.. .who had been the principal beneficiaries
in the colony of the crown's interventionism or who, like the large merchant houses of the capital, sought
to regain privileges the crown itself had abolished in the reforms of the late Bourbon era."
^"Coatsworth (1978,
colony's Creole
elite,
p.
95) notes "Mexican independence
carried out largely to separate
11
these institutions persist (see,
invested a lot in
human and
e.g.,
Acemoglu, 1995). For example, agents who have
physical capital will be in favor of spending
enforce property rights, while those
who have
less to lose
may
in the case of Mauritius, the presence of a middle-class that
not be.
had a stake
money
to
For instance,
in
democracy
and the rule of law seems to have been important.
Further evidence for the persistence of extractive state institutions into the indepen-
dence era
is
provided by the persistence of the most prominent extractive policies.
Latin America, the
full
panoply of monopolies and regulations, which had been created
by Spain, remained intact
sisted
and were even
sisal
for
most of the nineteenth century. Forced labor pohcies per-
intensified or re-introduced with the expansion of export agriculture
in the latter part of the
during the
In
boom
in
Slavery persisted in Brazil until 1886, and
nineteenth century.
Mexico, forced labor was reintroduced and persisted up to the
start of the revolution in 1910. Forced labor
was
also re-introduced in
Guatemala and El
Salvador to provide labor for coffee growing. In the Guatemalan case, forced labor lasted
until the creation of
democracy
independent African countries,
The
situation
is
in 1945.
for
example, by Mobutu
elites
was re-instated
in
many
in Zaire.
similar with other extractive policies. For example, marketing boards,
which were used by colonial powers
for the
Similarly, forced labor
for rent extraction,
were a ready made instrument
newly independent governments to tax agricultural producers (Bates, 1981). The
who
controlled the government
boards or from setting up more
had
little
to gain from abolishing these marketing
efficient institutions.
with the structure of political power
in the post
Very often these boards
independence
Nkrumah's support was primarily urban and amongst the
were rural and .Ashanti. So he did not find
restrict his taxation powers,
For example, in Ghana,
Fanti, while the cocoa farmers
beneficial to set
up
institutions that
would
and the marketing board provided a perfect instrument
raising taxes from a group that
Rimmer,
it
era.
fitted well
was not going to support him
politically in
1992). Fieldhouse (1999, p. 96) writes "once these boards
for
any case (see
came under
the control
of local politicians they were used to extract surplus from the rural producer, notionally
for
development purposes,
also argues (p.
in practice largely for
party and personal advantage."
146) that the fact there were no marketing boards
development experience of Australia.
12
was important
in
He
the
Institutions and Performance: Ordinary Least Squares Estimates
3
Data and Descriptive Statistics
3.1
Table
provides descriptive statistics for the key variables of interest.
1
for the
whole sample, and column
were ex-colonies and
GDP
for
per capita in 1995
Development Indicators
in the
all
first
settler mortality, institutions
is
and
GDP
The
data.
PPP adjusted and is taken from the World Bank's 1999 World
CD- Rom (a more detailed discussion of all data sources is provided
is
Income (GDP) per capita
ex-colonies in our sample
ago, income per capita today
is
a
had
will
be our measure of economic
relatively low levels of
income 400 years
(1.1 in
both
The
good measure of long run economic performance.
standard deviation of log per capita income both in the world sample and
sample are similar
column
for our base sample, limited to the 64 countries that
is
which we have
Appendix Table Al).
outcome. Since
2
The
cases),
in
our basic
indicatmg that the large income differences are
present in our sample.
We
use a variety of variables to capture institutional differences.
reported in the second row,
is
an index of protection against expropriation.
are from Political Risk Services (see,
first
used
m
e.g.,
Political Risk Services reports a value
between
corresponding to the highest expropriation
between 1985 and 1995 (values are missing
is
variable,
These data
Coplin, O'Leary, and Sealy 1993), and were
the economics and political science literatures by
against expropriation measure
Our main
risk.
for
and 10
We
many
Knack and Keefer
for each country
and
(1995).
year, with
use the average value for each country
countries before 1985). This protection
appropriate for our purposes since the focus here
differences in institutions originating
from
different types of states
and state
is
policies.
on
We
expect our notion of extractive state to correspond to a low value of this index, while
the tradition of rule of law and well enforced property rights should correspond to high
values.-'^
The next row
gives an alternative measure, coded from the Polity III data set
of
Gurr and associates (an update of Gurr 1997), which
in
1990 (see Appendix Table A4a). This variable
and
institutions, so
is
relevant for our purposes.
is
is
constraints on the executive
also closely related to state policies
To economize on
space, in the text
we
report only results with the protection against expropriation variable. Results using the
'^The protection against expropriation variable is most specifically for foreign investment, since Political
and Risk Services construct these data for foreign investors. However, as noted by Knack and Keefer
(1995), risk of expropriation of foreign and domestic investments are very highly correlated, and risk of
expropriation of foreign investment may be more comparable across countries. In any case, al! our results
hold also with a variety of other institutions and property rights measures (see Tables Al and A4a, b, c,
d, and e).
13
constraints on the executive and other measures are reported in the Appendix.
The next
The
first is
three rows give measures of early institutions from the
a measure of constraints on the executive in 1900 and the second
democracy
in 1900.
in 1900, so
we
row,
same Gurr data
This information
is
not available for countries that were
is
set.
an index of
still
colonies
assign these countries the lowest possible score in each index. In the third
we report an
alternative measure, constraints on the executive in the
first
year of
independence, which does not use this assumption, and assigns values for earlier years for
many
America countries and Neo-Europes that were already independent
of the Latin
the 19th-century.
in 1900,
which
is
The
final
in
row gives the fraction of the population of European descent
our measure of European settlement in the colonies, constructed from
McEvedy and Jones
and some additional sources;
(1978), Curtin et al (1995)
see
Appendix
Table A5).
The remaining columns
Table
in
give descriptive statistics for groups of countries
1
This
at different quartiles of the settler mortality distribution.
mortality
is
our instrument for institutions (this variable
is
useful since settler
described in more detail in the
is
next section).
Ordinary Least Squares Regressions
3.2
Table
2 reports
Ordinary Least Squares (OLS) regressions of log per capita income on the
The
protection against expropriation variable in a variety of samples.
linear regressions
are for the equation
logy,
where
y^
is
measure, X,
income per capita
is
in
fi
+ aR, + X-j + e^,
country
Rj
i,
a vector of other covariates, and
of interest throughout the paper
first
=
two columns of Table
is
is
e, is
(1)
the protection against expropriation
a
random
error term.
a, the effect of institutions on
2 are for the
The
coefHcient
income per capita. The
whole world sample, while the second two are
for
our former colonies sample. Since a priori we have no reason to believe that the relationship
between the institutions index and log
columns, we regress
GDP
GDP
per capita
per capita on a set of
of the distribution of the institutions index.
is
dummy
linear, in the
second and fourth
variables for different quartiles
The omitted group
the lowest quartile (with the worst institutions), and three
is
the set of countries in
dummies
for the other three
quartiles are included in the regression.^"
^'More generally,
it
may
therefore be
more appropriate
to interpret the
OLS and
I\'
estimates from
linear specifications as local average treatment effects giving the weighted average of the effects at different
14
Column
shows that
1
in the
whole world sample there
our measure of institutions and income per capita
means greater protection against expropriation
specification
is
dummies
appropriate, as the
gap between the various dummies
Columns
specification.
3-4
capita in our base sample
a strong correlation between
recall that the high value of the
risk.
Column
2 indicates that the linear
not very different from that implied by the linear
is
To
get a sense of magnitudes, let us
5.6,
measure
in this
while the 75th percentile of institutional measure
all
ex-colonies.
sample corresponds to a value of
is
The estimate
7A}^
column
in
indicates that there should be on average a 94 log point difference between the
3, 0.52,
log
percentile of the institutional
shows
1
compare the 25th
and 75th percentiles of the distribution of the institution variable among
The 25th
income per
of the institutions variable on
quite similar to that in the whole world, and Figure
this relationship diagrammatically.
index
are ranked in the expected order, and the
show that the impact
is
—
is
GDPs
of the corresponding countries (or approximately a difference equivalent to 154
percent of the
GDP
of the poorer country).
(approximately 540 percent). Therefore,
would imply a
would account
if
In practice, this
gap
is
185 log points
the effect estimated in Table 2 were causal,
fairly large effect of institutions
for
GDP
it
on performance: differences in institutions
approximately a quarter of the income per capita differences across
countries.
Many
social scientists, including
and coauthors have argued
Montesquieu [1784] (1989), Diamond (1997) and Sachs
for a direct effect of climate
Mellinger, and Sachs (1998) and Hall and Jones (1999)
on performance, and Gallup,
document the
distance from the equator (latitude) and economic performance.
columns
5-8,
we add the absolute value
To
correlation between
control for this, in
of the distance from the equator in degrees (lati-
tude) as a regressor (we follow the literature in using a measure of latitude that
between
and
1).
This changes the coefficient of the index of institutions
itself is also significant
column
and has the sign found by previous
7 indicates that a ten degree (0.10) distance
between Mauritania and Morocco,
in
is
America
nificant, the continent
8,
we
also
as the omitted group.
dummies
The
Latitude
coefficient 1.60 in
gap from the equator, such as the gap
add dummies
dummy
for Africa,
Although expropriation
are also statistically
example, the coefficient of the Africa
in
and quantitatively
column
Asia and other
risk
remains
significant.
8 indicates that in our
points (as suggested in Angrist and Imbens, 1995).
'^We
scaled
associated with approximately 16 percent difference
income per capita. In columns 6 and
continents, with
studies.
little.
is
focus on the inter-quartile range in order to reduce the influence of measurement error.
15
sig-
For
sample
African countries are 90 log points (approximately 145 percent) poorer even after taking
the effect of institutions into account.
Overall, the results in Table 2
show a strong correlation between
institutions
economic performance. Nevertheless, there are a number of important reasons
for
and
not in-
terpreting this relationship as causal. First, our measure refers to current institutions, so
it is
plausible that rich economies are able to afford, or perhaps prefer, better institutions.
Arguably more important than
determinants of income differences that
nally, the
problem, there are
this reverse causality
will naturally
be correlated with institutions. Fi-
measures of institutions are constructed ex post, and the analysts
had a natural
bias in seeing better institutions in richer places.
introducing positive bias in the
OLS
As
well as these
downwards.
All of these
problems could be solved
we
is
this variable in
a plausible instrument.
more
Curtin.
problems
The next
estimates
for institu-
accounting for the institutional
Our
discussion
by the settlers during the time of
section will describe the construction of
Data on the Mortality of Early Settlers
the mortality of European settlers
books and
In a series of
come
largely from the
articles over a period of nearly
a variety of sources to document the mortality of Europeans in
work of Philip
40 years, Curtin has used
new
colonies. Systematic
military medical record keeping began only after 1815, as an attempt to understand
so
many
soldiers were dying in
and dealt with British
forces
some
The
places.
first
why
detailed studies were retrospective
between 1817 and 1836. The U.S. and French governments
quickly adopted similar methods (Curtin 1989, p.
also available for the
is
detail.
4
Our data on
in
observe, but not have a direct effect on performance.
in Section 2 suggests that the mortality rates faced
colonization
OLS
we had an instrument
if
Such an instrument must be an important factor
variation that
may have
estimates, the fact that the institutions variable
measured with considerable error creates attenuation and may bias the
tions.
many omitted
Dutch East
Indies.
By the
3
and
1870s,
p.
5),
and some early data are
most European countries published
regular reports on the health of their soldiers.
The standard measure
reports the death rate
is
annualized deaths per thousand
among
1,000 soldiers where each death
mean
is
strength. This
replaced with a
measure
new
soldier.
Curtin (1989 and 1998) reviews in detail the construction of these estimates for particular
places and campaigns, and assesses which data should be considered reliable.
The data can be divided
into two parts.
16
Curtin (1989), Death by Migration^ deals
primarily with the mortahty of European troops from 1817 to 1848. At this time modern
medicine was
how
still
and none of the European
in its infancy,
to design their
campaigns so as to reduce mortality. This period
of death from disease for both soldiers
main cause
of death
is
well before the
was understood, and these were the major causes
control of malaria and yellow fever
(the third
militaries yet understood
and
where mortality was high
settlers in places
was gastrointestinal
These mortality rates can
diseases).
therefore be interpreted as reasonable estimates of settler mortality.
with substantial evidence from other sources
They
are consistent
example, Curtin 1964 and Curtin
(see, for
1968).
Curtin (1998), Disease and Empire, adds similar data on the mortality of soldiers
the second half of the nineteenth century.
These numbers have
to
in
be used with more care,
because there was a growing awareness of how to avoid epidemics of the worst tropical
diseases, at least during short military
at the
campaigns. For example, the campaign
end of the nineteenth century had very low mortality rates because
and well-managed
(see
Table
A2 and Figure
1).
campaign certainly underestimate the mortality
Ethiopia,
we did not exclude
hypothesis. In
that this
is
all cases,
we use the
it
Ethiopia
was short
Although the mortality rates from
successful
this
in
rates faced by the settlers in
country because excluding
earliest available
this
number
it
would have helped our
for each country, reasoning
the best estimate of the mortality that settlers would have faced, at least until
the 20th century.
Appendix B reviews
in detail
how our data
assumptions that check the robustness of our
for
are constructed,
results.
and describes alternative
The main gap
in the
Curtin data
is
South America since the Spanish and Portuguese militaries did not keep good records of
mortality.
However, Gutierrez (1986) used Vatican records to construct estimates
mortality rates of bishops in Latin America from 1604 to 1876.
for the
Because these data overlap
with the Curtin estimates for several countries, we are able to construct a data series for
South America
(details are in the Appendix).^''
Curtin (1964) also provides estimates of
mortality in naval squadrons for different regions which
We
estimates of mortality in South America.
methods produce remarkably similar
Nevertheless, since
we
in
to generate alternative
Table 5 that these alternative
results.
^"Combining data from a variety of sources
settler mortality.
show
we can use
will surely
introduce measurement error in our estimates of
are using settler mortality as an instrument, this
error does not lead to inconsistent estimates of the effect of institutions on performance.
17
measurement
Institutions and Performance: Instrumental Variables Results
5
Determinants of Current Institutions
5.1
Mathematically, our theory can be expressed in the following way.
we
(1),
which
rewrite here,
=
log y^
tI
+ aR^ + XIj +
£,-,
describes the relationship between current institutions and log
R
is
C
is
European settlements
and
we have
In addition
+ PrQ + X,'7fi +
un,
(2)
Q=
Xc
+ PcS^ + x'ac +
yci
(3)
+
(4)
=
As
+
13s
log
M, +
X[-is
us^
the measure of current institutions (protection against expropriation between
1985 and 1995),
1900),
GDP.
Afi
5,
where
(1)
=
R^
M
is
our measure of early (circa 1900) institutions,
in the
S
is
the measure of
colony (fraction of the population with European descent in
X
mortality rates faced by settlers.
is
a vector of covariates that affect
variables.
all
"
The
simplest identification strategy
equation
with
£j
quality
(1).
— even
likely to
is
to use log Af, directly as an instrument for
This identification strategy
if
will
C, and S^ were correlated with
be valid as long as logM,
e,.
For example,
migrate to places with better resources and
still
had an
effect
soil quality,
Nevertheless, our identification strategy of using
valid.
We
if
logM;
start with this identification strategy,
/2,
in
uncorrelated
is
Europeans were more
and
if
resources and soil
on income, there would be a correlation between
would make European migration patterns an invalid instrument
be
Equation
S^
and
e,.
This
for current institutions.
directly as an instrument
would
and then use the other equations to
derive overidentifying restrictions.
Figure 2 illustrates the relationship between the (potential) settler mortality rates and
the index of institutions.
We
use the logarithm of the settler mortality rates, since there
are no theoretical reasons to prefer the level as a determinant of institutions rather than
the log, and using the log ensures that the extreme African mortality rates do not play a
disproportionate
role.
As
log of the mortality rate
it
happens, there
and our measure of
is
an almost linear relationship between the
institutions.^^
^^Note that Gambia has a surprisingly good score for protection against expropriation
exclude this country, since doing so would help our hypothesis.
18
risk.
We
do not,
.
In Table 3,
we document that
esized in Section
(4).
In the top panel,
Column
variables.
we present OLS
In particular,
2.
1
we
works through the channels hypoth-
this relationship
regressions of equations (2), (3), and
regress the protection against expropriation variable on the other
uses constraints faced by the executive in 1900 as the regressor, and
shows a close association between early institutions and institutions today. For example,
past institutions alone explain 21 percent of the variation in the index of current institutions.
The second column adds
Columns
3
the latitude variable, with
and 4 use the democracy index, and confirm the
little effect
on the estimate.
results using the constraints
on the executive variable.
Both constraints on the executive and democracy
indices assign low scores to countries
that were colonies in 1900, and do not use the earliest post-independence information for
Latin American countries and Neo-Europes
columns
In
.
approach and use the constraints on the executive
control separately for time since independence.
5
and
6,
we adopt an
in the first year of
The
alternative
independence and also
results are similar,
and indicate that
early institutions persist (and time since independence has a positive effect on institutions
today)
Columns
7 and 8
show the association between protection against expropriation and
European settlements.
The
29 percent of the variation
in
fraction of
Europeans
in
1900 alone explains approximately
our institutions variable today. Columns 9 and 10 show the
relationship between the protection against expropriation variable and the mortality rates
faced by settlers.
This specification
These regressions show that
in institutions
Panel
B
we observe
will
be the
first-stage for
settler mortality alone explains 26 percent of the differences
today.
of Table 3 provides evidence in support of the hypothesis that early institu-
tions were shaped, at least in part, by settlements,
mortality.
Columns
democracy
in
ulation of
our main 2SLS estimates.
1-2
and
-5-6
relate our
and that settlements were affected by
measure of constraint on the executive and
1900 to the measure of European settlements in 1900 (fraction of the pop-
European decent). Columns 3-4 and 7-8
rates of settlers.
relate the
same
variables to mortality
These regressions show that settlement patterns explain around 50 per-
cent of the variation in early institutions.
between settlements and mortality
Finally,
columns
9
and 10 show the relationship
rates.
Overall, Table 3 provides support for our hypothesis that mortality rates faced by
potential settlers affected settlement behavior, and that settlement behavior affected the
type of institutions that the colonialists brought to the country, and finally that early
19
in-
stitutions persisted to the present.
for current institutions,
We
next use settler mortality directly as an instrument
and look at the impact of current institutions on performance.
Performance
5.2
Institutions and Economic
The
basic results are presented in Table 4.
protection against expropriation variable,
R,
where M,
is
not appear
=
C,
+
tion (1)
+ X[5 +
So we estimate equations
of Table 4 reports
and Panel
B
tutions on income per capita
in fact larger
is
0.94.
of the coefficient of interest,
This estimate
than the
that this variable does
with two stage least squares
Column
OLS
1
a from equa-
displays the strong
and current institutions
The corresponding 2SLS estimate
is
as
(5)
(5) jointly
(log) settler mortality
3.
it
treat the
/?,.
2SLS estimates
between
sample, also shown in Table
and
and
We
(1).
v,
gives the corresponding first-stages.^^
first-stage relationship
error of 0.16,
(1)
is
and model
our settler mortality rate. The exclusion restriction
in (1).
A
as endogenous,
i?,,
l3\ogM,
(2SLS), using logM, as an instrument for
Panel
The main equation
is
in
our base
of the impact of insti-
highly significant with a standard
estimates reported in Table
2.
This suggests
that measurement error in the institutions variables that creates attenuation bias
to be
more important than
reverse causality
To evaluate the quantitative
and omitted variables
significance of these estimates,
biases.
is
likely
^^
we again compare the 25th
and 75th percentiles of the distribution of the protection against expropriation variable
among
all
ex-colonies.
The
baseline
2SLS estimate
of 0.94 implies that there should be a
171 log point (450 percent) diff'erence between the average
GDPs
of the countries at the
25th and 75th percentiles of the distribution of protection against expropriation, as com-
pared to the 185 log points (540 percent) difference in the data. Therefore, these estimates
suggest that over three-quarters of the income per capita diff'erences across countries
to differences in institutions.
is
due
Figure 3 plots the predicted values of income per capita from
our baseline regression, which includes only the institutions index as a regressor, against
actual income per capita for the countries in our sample (as well as the 45 degree line).
'®We have
morfrom the same disease environment. This clustering has little effect on the standard
errors, and does not change our results.
^'^For example, the fact that Gambia has such a high score in the protection against expropriation index
(see Figure 2) suggests that there is considerable measurement error in this variable. See also footnote 20
for further discussion of the measurement error problem.
also run these regressions with standard errors corrected for possible clustering of the
tality rates assigned
20
Most observations are
which confirms the explanatory power
close to the 45 degree line,
of our regression.^*
Column
coefficient
shows that adding latitude does not change the relationship; the institutions
2
now
is
1.00 with a standard error of 0.22.^^
now has the "wrong"
studies
may
because
sign
and
insignificant.
is
Remarkably, the latitude variable
This result suggests that
many
previous
have found latitude to be a significant determinant of economic performance
it is
correlated with institutions (or with the exogenous
component of institutions
caused by early colonial experience).
Columns
and 4 document that our
3
results are not driven
we exclude
the U.S., Canada, Australia, and
significant,
and
now
New
by the Neo-Europes.
When
Zealand, the estimates remain highly
For example, the coefficient for institutions
in fact increase a little.
and
1.28 (s.e.=0.36) without the latitude control,
1.2 (s.e.=0.35)
when we
is
control for
latitude.
Columns
5
and
6
show that the
result
is
also robust to
The estimates without Africa
from our sample.
are
precise.
For example, the coefficient for institutions
control,
and
In
still
0.58 (s.e.=0.12)
Columns 7 and
8,
and other, with America
change the estimated
5 percent level,
The
as the omitted group).
effect of institutions,
fact that the African
is
dummy
dummy
smaller, but also
more
0.58 (s.e.=0.1) without the latitude
to the regressions (for Africa, Asia
The addition
of these
dummies does not
is
significantly different
at the
from that of America.
insignificant suggests that the reason
is
why
African
not due to cultural or geographic factors, but mostly accounted for
by the existence of worse institutions
In
the African countries
and the dummies are jointly insignificant
Asia
for
somewhat
all
control for latitude.
we add continent dummies
though the
countries are poorer
when we
is
dropping
Appendix Table A4a,
b, c, d,
in Africa.
and
of alternative measures of institutions.
e,
we repeat the same
exercise using a variety
These are constraints on the executive from the
Polity III dataset, an index of law and order tradition from Political Risk Services, a
measure of property rights from the Heritage Foundation, a measure of rule of law from
the Fraser Institute, and the efficiency of the judiciary from Business International (see
Tables
Al and A4
for
more
to those reported in Table 4.
details).
The
results
and the magnitudes are very similar
Furthermore, in Appendix Table A3, we also report very
^^The cloud of points in this graph indicates a slope somewhat smaller than the 45 degree line, so our
2SLS estimates may be underpredicting the income level in the countries with the worst institutions.
•'^In 2SLS estimation, all covariates that are included in the second stage, such as latitude, are also
included in the
first
stage.
we do not report them
When
these
in the tables to
first
stage effects are of no major significance for our argument,
save space.
21
similar results with 1970 values for the constraints on the executive
in 1970,
which show that the relationship between institutional measures and income per
capita holds across time periods.
We
and income per capita
^°
should note at this point that
if
we
sample to African countries
limit the
only,
the first-stage relationship using the protection against expropriation variable becomes
considerably weaker, and the 2SLS effect of institutions
2SLS
is
no longer
The
significant.
be significant when we use the constraint on the
effect of institutions continue to
executive variable or Fraser Institute's rule of law variable (and not with other measures).
we conclude that the
Therefore,
relationship between settler mortality and institutions
is
weaker within Africa.
show a large
Overall, the results in Table 4
mance. In the
rest of the paper,
we
investigate the robustness of these results.
Robustness
6
Alternative samples
6.1
In Table
4
on economic perfor-
effect of institutions
5,
we
we reported
investigate the robustness of our results in different subsamples. In Table
results in samples
whether the results are robust
which differed by region.
to varying degrees of
constructing the mortality estimates.
(1989), Death by Migration, which
In this smaller sample,
we
find
is
In
columns
In
columns
.3
4,
2,
and
in
and
always to pre-1840 data.
refers
Table
in the first stage
when
we add data from the second book by Curtin
similar to those reported in Table
information (described
in detail in
4.
latitude
is
is
0.97
included
4).
(1998), Disease
and
This increases the sample to 30
and reduces both the estimate and the standard error a
now much more
and the
on income per capita
1.1 (s.e.=0.33)
Empire, for the earliest date available for each country.
countries,
to see
we only use data from Curtin
effect of institutions
(compared to 0.94 and 1.00
and
and
an even stronger relationship both
(s.e.=0.22) without controlling for latitude,
is
data quality and different methods of
for 17 countries
second stage. The 2SLS estimate of the
in the regression
1
Here our objective
In
columns
little;
5
and
the results are
6,
we use other
the Appendix), to assign mortality rates to neighboring
'"The presence of alternative measures of institutions also enables us to ascertain whether the difference
between OLS and 2SLS estimates could really be due to measurement error in the institution variable. If
we use the constraints on the executive measure as an instrument for the protection against expropriation
measure, this would solve the measurement error, but not the endogeneity problem. This exercise leads
to an estimate of the effect of protection against expropriation equal to .87 (with standard error ,16),
which suggests that measurement error in the institutions variables is very important, and can explain
tlie whole difference between the OLS and 2SLS estimates.
22
countries with the
same
the estimates Httle.
In
columns 7 and
columns 9 and
In
10,
we use an
we add data
8,
The
from naval stations instead of bishops.
4.
This increases the sample to 45, and changes
disease environment.
from Sierra Leone
the
is
America, but this time
results are identical to those obtained in Table
alternative ("conservative") coding of the African data.
In particular, rather than using African estimates
rates to areas that shared the
for Latin
same
from small samples, we assign mortality
For example, the same number
disease environment.
applied to Benin, Cote d'lvoire, Ghana, Guinea, Nigeria, Togo, and
number from French Soudan
similar to our baseline results.
is
Finally, in
home
mortality rates relative to
assigned to Mali and Niger. This leads to estimates very
columns 11 and
12,
we report
regressions using
country, which again yield very similar effects.
Overall
conclude that the results hold in different samples with varying degrees of data
reliability.
Additional controls
6.2
The
validity of our
2SLS
tality in the past has
Table 4 depends on the assumption that settler mor-
results in
no direct
effect
on current economic performance.
presumption appears reasonable, here we substantiate
for
we
many
Overall,
we
the inclusion of these variables, and
find that our results
many
insignificant once the effect of institutions
further by directly controlling
change remarkably
variables emphasized in previous
is
controlled
little
with
work become
for.
Porta, Lopez-de-Silanes, Shleifer and Vishny (1999) argue for the importance of
colonial origin (main colonizer) as a determinant of current institutions.
the colonial power could also matter because
argued by Landes (1998).
colonizer (for Belgium,
In
column
1,
it
might have an
we add dummies
Germany, the Netherlands,
Britain as the omitted group). ^^
The top panel
Italy,
effect
the protection against expropriation variable.
dummies
for the identity of the
2SLS
^^In the table,
OLS
income per capita
Interestingly, the first-stage estimates
by the colonizers
is
coefficient
(i.e.,
is
23
on
latitude.
robust to
show that these
a p- value over
.15),
when the
controlled for (even though the
dummy
we only report the coefficients on the Belgian, French and Spanish dummies
dummies are also included in the estimation.
space, but the other
main
estimates, the second
The even numbered columns add
are jointly insignificant at the 15 percent level
effect of mortality rates faced
identity of
France, Spain, Portugal, with
gives the
or without latitude, the effect of institutions on
the inclusion of these dummies.
The
through culture, as
panel gives the first-stage, and the bottom panel gives the corresponding
With
this
of the variables that could plausibly be correlated with both settler mortality
and economic outcomes.
La
it
Although
to save
which captures the Congo,
for being a Belgian colony,
show
results therefore
institutions once
we
These
significantly negative).
evidence for a major influence of the identity of the colonizer on
little
control for the effect of the mortality rates faced by colonialists.
common
therefore contrast with a
The reason why
better. ^^
is
They
perception that the British former colonies perform
British colonies appear to perform better in other studies seems
to be that Britain colonized places
where settlements were possible, and
this
made
British
colonies inherit better institutions (in our sample too, ex-British colonies have considerably
better institutions, so
dummy
if
we drop the
log mortality variable from the regression, the British
has a significant and large positive
on institutions). ^'^
eff'ect
Hayek (1960) and La Porta, Lopez-de-Silanes,
size the
importance of
our sample,
dummy
all
columns 3 and
4,
and Vishny (1999)
we
also
empha-
control for legal origin.
countries have either French or British legal origins, so
In
we simply add
a
French legal origin (many countries that are not French colonies nonetheless
for
Once
have French legal origin).
per capita
In
legal origin.
Shleifer
again, our estimate of the effect of institutions on
income
is unaff'ected.'^''
An argument
dating back to
performance. To control for
Max Weber views
this, in
columns
5
religion as a key
and
6,
we add the
determinant of economic
fraction of the populations
that are Catholic, Muslim, and of other religions, with Protestants as the ommitted group.
In the table
these
we report the
dummies
significance level (p-value) of the corresponding F-statistic for
as well as the
2SLS estimate
of the effect of institutions. Religion appears
^^2SLS estimates show that the identity of the colonizer does have an effect on income per capita (i.e.,
effect not working through institutions). In particular, Belgian and Spanish ex-colonies do better than
expected relative to their institutions. This is because Belgian and Spanish ex-colonies have substantially
worse institutions, but the performance of these countries is not as bad as would be expected on the
basis of their institutions alone.
The Belgian and Spanish dummies therefore take on positive values
in the 2SLS estimatation to compensate for this.
For example, in the case of Belgian colonies, the
first-stage effect is -2.9, which multiplied with the coefficient of 1.1 on institutions, implies that Belgian
ex-colonies should be approximately 320 log points poorer. The positive coefficient of 2.4 in the secondstage somewhat compensates for this, and leaves a net effect of 240-320=-80 log points (122 percent) for
an
Belgian colonies
—
^^In addition,
if
colonies,
we
still
a very large
we run our
basic
effect!
IV regression only
for the subset of countries that
were former British
For example, without latitude, the coefficient of the average
obtain very similar results.
protection against expropriation variable
is
1.06
(s.e.
= .24); and
with latitude,
it
is
1.33 (s.e.=.52).
In
contrast, in the sample of non-British colonies, these coefficients are 1.18 (s.e.=.46) and 1.09 (s.e.=.45),
and not
statistically different
^''This time, however,
we
we
from those
in the British colonies
find that French legal origin
control for the mortality rates faced by the settlers.
of institutions on performance
is
is
But
only sample.
associated with worse institutions, even
when
second stage estimation, when the
effect
in the
taken into account, the French legal origin dummy is positive. In fact,
is actually positive: -67x1. 1-1- 89=- 15 log points (approximately
the net effect of having French legal origin
15 percent).
This suggests that although countries with a French legal origin have worse institutions,
they are somewhat richer because of other factors (uncorrelated with legal origin and settler mortality).
24
highly correlated with income per capita (in particular, higher fractions of Catholics and
Protestants are associated with greater income per capita), but the 2SLS estimate of the
effect of institutions
hardly changes. Finally, column 7 adds
simultaneously. Again, these controls have very
Another concern
ease environment.
is
Our instrument may
variables. In Table 7,
erage,
that settler mortality
we add a
set of
little effect
correlated with climate and current dis-
is
therefore be picking
(all
up the
direct effect of these
humidity. In the table
maximum
we report
data from Parker, 1997). Again, they have
humidity, and afternoon
joint significance levels for these
little effect
on our estimates.
Sachs and a series of coauthors have argued for the importance of malaria
African poverty
Gallup
(see, for
et al., 1998).
estimate
may be
explaining
in
example. Bloom and Sachs, 1998, Gallup and Sachs, 1998, and
Since malaria was one of the main causes of settler mortality, our
capturing some of the direct effects of malaria.
argument since the prevalence of malaria
highly endogenous;
is
We
it is
are skeptical of this
the poorer countries
with worse institutions that have been unable to eradicate malaria."^ Moreover,
to imagine
are: av-
minimum and maximum
high temperatures, and
monthly low temperatures, and morning minimum and
variables
on our main estimate.
temperature and humidity variables. These
minimum and maximum monthly
minimum and maximum
the variables in this table
all
how malaria could have such
it is
hard
a large effect on economic performance. Although
Sachs and coauthors argue that this works through poor health, high mortality and absenteeism, most people
disease
—
fatal to
if
who
live in
high malaria areas have developed some immunity to the
they survive to the age of
them
(see Curtin, 1998, or
will
have very high social
they can
fall
Bruce-Chwatt, 1980).
malaria should not have a very large
it
five,
costs).
effect
periodically but malaria
ill
We
might therefore expect that
In contrast, for Europeans, or anyone else
is
fatal, ^^
malaria a key determinant of whether Europeans could settle
second part of Table 7 by controlling for the fraction of the population
"^For example, the U.S. eliminated malaria from the
from Queensland
(see Crosby, 1986, pp. 141-142).
Panama Canal
Even
in Africa,
making variations
We
who
do
this in the
live in
local.
It is
an area
zone, and Australia eliminated
there are very successful campaign
against malaria, including those in Algeria and that conducted by the Rio-Tinto Zinc mining
Zambia (then Northern Rhodesia).
-^Some types of malaria are quite
who has
in a colony.
In any case, controlling for malaria does not change our results.
it
not
on economic performance (though, obviously,
not been exposed to malaria as a very young child, malaria
in
is
company
therefore quite possible for a person to have
to the local version of malaria, but to be highly vulnerable to malaria a short distance away.
in
immunity
This
is
probably one explanation why Africans had such high mortality when they were forced to move by colonial
powers. For example, African labor on the Congo-Ocean railroad had mortality of 240 per thousand at
its peak, and 100 per thousand on average (Curtin, et al., 1995).
25
where falciporum malaria
In
1998).
is
endemic
in
1994 (as constructed and used by Gallup
Appendix C, we show that the inclusion
et al,
of an endogenous variable positively
correlated with income or institutions will bias the coefficient on institutions downwards.
Since the prevalence of malaria in 1994
will therefore
coefficient
Table
underestimate the
is
highly endogenous, controlling for
on protection against expropriation
7, .69
instead of .94 as in Table
a standard error of
.25,
4.
is
is
current disease environment, but likely through
A
related concern
is
is
not through
its effect
Hence, we conclude
its
that in colonies where Europeans settled, the current population
One might be
the direct effect of having more Europeans.
To
population of European descent
and Jones (1978), Curtin
worried that we are capturing
control for this
we add the
et al (1995)
is
and some additional sources; see Appendix Table
insignificant, while the effect of institutions
highly significant, with a coefficient of .96 and a standard error of
7,
we add
all
descent, together,
income per capita
In
.28.
affected
Finally, in Table 8
effect of protection against
remains
column
these variables, for temperature, humidity, malaria, and
and the estimate of the
is
fraction of the
columns 7 and 8 of Table 7 (constructed from McEvedy
in
Interestingly, this variable
Table
correlation with the
on institutions.
consists of a higher fraction of Europeans.
A5).
result, the
remains highly significant with
it
highly insignificant.^^
that the effect of settler mortality on performance
directly
estimated to be somewhat smaller in
Nevertheless,
while malaria itself
As a
on performance.
effect of institutions
it
9 of
European
expropriation on
little.
we add a
variety of controls for other potential determinants of
income per capita. These include a measure of ethnolinguistic fragmentation, measures of
soil
and natural resources, and whether the country
quality
in the
Appendix
for definitions). For
example, in column
1
is
landlocked (see the Table
where we include ethnolinguistic
fragmentation, the coefficient of protection against expropriation
is
Al
is
0.74 (s.e.= 0.13), which
only slightly smaller than our baseline estimate. Since ethnolinguistic fragmentation
likely to
be endogenous with respect to development
(i.e.,
is
ethnolinguistic fragmentation
tends to disappear during periods of growth and formation of centralized markets, see
Weber, 1976, or Andersen, 1983) and
is
correlated with settler mortality, the estimate of
0.74 likely understates the effect of institutions on income (see
also
change
"^Malaria
descent,
of
and
is
little
when
Appendix
controls for natural resources, soil quality
in column 9 when we control
and humidity variables. Nevertheless,
marginally significant
eight temperature
them with malaria, are always
jointly insignificant.
are insignificant has a p-value of .39.
26
The estimates
and being landlocked
for latitude,
all
C).
fraction of
European
of these variables, or any subset
For example, the F-test that
all
of these variables
We
are added, and the controls themselves are insignificant.
effect of variations in institutions
and
therefore conclude that the
caused by early colonial experience on income
likely captures the causal effect of institutions
and government
policies
is
robust,
on economic
well-being.
OVERIDENTIFICATION TESTS
6.3
We can also investigate the validity of our approach by using overidentification tests.
Recall
that according to our theory, settler mortality affected settlements; settlements affected
early institutions;
and early
institutions affected current institutions. Mathematically,
expressed these links by equations
variables, C, S,
C
and S
and
M,
(2),
and
(3),
(4).
We
can
test
has a direct effect on income per capita, log
as additional instruments.
The
overidentification test
we
whether any of these
y,
by using measures of
would then check whether
the 2SLS estimates change significantly once additional instruments are included.
The
1.
2.
overidentification test will reject the validity of our approach
the equation of interest (1) does not have a constant coefficient,
aiRi
+ Ei,
C
5
or
with
3.
if:
logy^
=
At
+
or
has a direct effect on income per capita, logy,
£i),
i.e.,
(i.e.,
either
5
or
C
is
correlated
or
settler mortality,
M,
has an
effect
on
logy,-
that works through another variable,
such as culture.
The
overidentifying restrictions implied by our approach are never rejected. This im-
plies that, subject to the caveats related to the
can rule out
all
three of the above possibilities.
settler mortality is a valid
power of the overidentification
instrument and that we are estimating the
The
some other variable correlated with
expropriation on
while Panel
B
and related
we report the 2SLS estimates
(i.e.,
not capturing the
results, are reported in
Table
9.
of the effect of protection against
GDP per capita using a variety of instruments other than mortality rates,
gives the corresponding first-stages.
to those reported in Table 4.
in
effect of institutions
settler mortality).
results of the overidentification tests,
In the top panel. Panel A,
we
This gives us additional confidence that
on current performance with our instrumental variable strategy
effect of
tests,
For example,
in
1900 as the only instrument for institutions.
27
These estimates are always quite close
column
1,
we use European settlements
This results
in
an estimated
effect of .90
—
(with standard error .13), as compared to our baseline estimate of
add
latitude,
and use other instruments such as constraint on the executive
D
reports an easy-to-interpret version of the overidentification test.
the log of mortality as an exogenous regressor.
a direct effect on income per capita,
and
In
significant.
column
1,
1900 and
in
independence, and democracy in 1900.
in the first year of
Panel
The other columns
.94.
all
cases,
it
is
If
It
adds
mortality rates faced by settlers had
we would expect
come
this variable to
small and statistically insignificant.
in negative
For example, in
log mortality has a coefiicient of -.07 (with standard error .17). This confirms
that the impact of mortality rates faced by settlers likely works through their
eflfect
on
institutions.
Finally, for completeness, in Panel
C we
report the p-value from the appropriate x^
2SLS
overidentification test. This tests whether the
ments indicated
in
Panels
A
and
B
coefficients estimated with the instru-
versus the coefficients estimated using (log) settler
mortality in addition to the other instruments are significantly different
column, the coefficient using European settlements alone
European settlements and
is
log mortality as instruments).
compared
We
(e.g., in
the
first
to the estimate using
never reject the hypothesis
that they are equal; so these results also show no evidence that mortality rates faced by settlers
have a direct
effect
— or an
effect
working through a variable other than institutions
on income per capita.
7
Many
Concluding Remarks
economists and social scientists believe that differences in institutions and state
policies are at the root of large differences in
little
income per capita across countries. There
is
agreement, however, about what determines institutions and the attitude of govern-
ments towards economic progress. This makes
it
difficult to isolate
exogenous sources of
variation in institutions to estimate their effect on performance. In this paper
we argued
that differences in colonial experience could be a viable source of exogenous differences in
institutions.
Our argument
rests
on the following premises. Europeans adopted very different col-
onization strategies, with different associated institutions. In one extreme, as in the case
of the U.S., Australia and
New
Zealand, they went and settled in the colonies and set-up
institutions that enforced the rule of law
and encouraged investment.
In the other ex-
treme, as in the Congo or the Gold Coast, they set up extractive states with the intention
of transferring resources rapidly to the metropole.
28
The
slave trade
is
perhaps the most
extreme example of
These
this behavior.
sets of institutions
were very detrimental to
in-
vestment and economic progress. Which colonization strategy was adopted was, at least
determined by the
in part,
feasibility of
European settlement.
where Europeans
In places
faced very high mortality rates, they could not go and settle, and they were
up extractive
to set
states. Finally,
present. Determinants of whether
have an important
effect
we argue that
likely
these early institutions persisted to the
Europeans could go and
We
on institutions today.
more
settle in the colonies, therefore,
exploit these differences as a source of
exogenous variation to estimate the influence of institutions on economic performance.
We
document these hypotheses
in the
data by showing a high correlation between
mortality rates faced by soldiers, bishops, and sailors in the colonies and European settlements; between European settlements and early measures of institutions; and between
early institutions
and institutions today.
come per capita using
We
estimate large effects of institutions on in-
Our estimates imply
this source of variation.
income per capita.
institutions account for roughly three-quarters of the differences in
also
document that
this relationship
is
that differences in
not driven by outliers, and
is
We
robust to control-
ling for climate, current disease environment, religion, natural resources,
and current race
composition.
It
is
useful to point out that our findings
do not imply that institutions today are
We
predetermined by colonial policies and cannot be changed.
rience as one of the
settlers are
many
factors affecting institutions.
emphasize colonial expe-
Since mortality rates faced by
arguably exogenous, they are useful as an instrument to isolate the
institutions on performance.
In fact, our reading
economic gains from improving
institutions, for
is
effect of
that these results suggest substantial
example as
in the case of
Japan during
the Meiji restoration or South Korea during the 1960s.
There are many questions that our analysis does not address.
to
some degree
as a "black-box":
(or increasing other
our results indicate that reducing expropriation risk
measures of propoerty rights enforcement) would
gains in income per capita, but do not point out
improvement
to an
risk,
Institutions are treated
in these institutions.
result in significant
what type of concrete
Institutional features, such as expropriation
property rights enforcement or rule of law, should probably be interpreted as an
equilibrium outcome, related to some more fundamental ''institutions",
vs.
steps would lead
parliamentary system, or the presence of
directly.
rights
.A.
more
detailed analysis of the effect of
and expropriation
risk
is
common
29
presidential
law, etc., which can be changed
more fundamental
an important area
e.g.,
institutions on property
for future study.
Appendix A:Data Sources
8
Appendix Table Al shows the complete
their definitions
9
list
of variables that
we
use, together
with
and sources.
Appendix B:Construction of Settler Mortality Data
Appendix Table A2 reports our estimates
The
for settler mortality in each country.
first
column of mortality estimates shows data from Curtin (1989)
for countries located in the
exact place for which we have a mortality estimate from the
first
These data are used
century.
We
in
have direct estimates of
Guyana
the
first
two columns of Table
half of the nineteenth
5.
Canada
settler mortality for Algeria (1831-38),
(1819-36), Jamaica (1817-36), Malta (1817-36), Mauritius (1818-36),
We
(1846-55), Senegal (1819-38), and Sierra Leone (1819-36).
has been a simple change of name, but the country
is
the
New
also have
estimates for Britain (1830-36) and France (1820-22 and 1824-26).
In
(1817-36),
some
Zealand
comparable
cases there
same geographically.
We
can
therefore use the estimate of settler mortality from Ceylon (1817-36) for Sri Lanka, from
Coastal
Burma
(1829-38) for
Myanmar, from Cape Colony (1818-36)
and from the Windwards and Leewards islands
and Tobago.
for
It is also
for
South Africa,
(1817-36) for Barbados and for Trinidad
reasonable to use the mortality estimate from Bengal (1830-38)
Bangladesh, from the Dutch East Indies (1819-28) for Indonesia, and from the Straits
Settlements (1829-38) for Malaysia and Singapore. Note that there
is
a large variation in
mortality between the Straits Settlements and Dutch East Indies, even though these areas
are quite close geographically. This
is
because there exists substantial variation in disease
environment, particularly for malaria, even
in
neighboring areas.
For example, Curtin
(1998) explains in detail the substantial variation in malaria risk just within Madagascar
at the end of the nineteenth century.
and the precise vegetation, play a
Micro-climate, particularly the pattern of rainfall
significant role in determining the risk of malaria. See
Curtin (1989) for further discussion.
For large countries where we have more than one regional estimate, we need to make
a choice.
For the U.S. we use the estimate for the Northern United States (1829-38).
For India we use the estimate from Madras (1829-38); this
higher than for Bombay.
None
is
lower than for Bengal but
of our regression results are significantly affected
if
we use
alternative regional data for these countries.
The second column
of mortality estimates in
30
Appendix Table A2 adds data from Curtin
Whenever Curtin provides
(1998), which covers the second half of the nineteenth century.
more than one estimate, we use the earhest
available
number
We
for each country.
mortality data directly from Mexico in 1862-63 and Tunisia in 1881.
As we note
have
in the
data from the second half of the nineteenth century have to be used with care, because
text,
there were improvements in the practice of military medicine around the mid-nineteenth
century, but
most of this would not have helped to reduce the mortality of civilian
settlers.
For example, the British campaigns of 1882 in Egypt, 1867-68 in Ethiopia and the 1884-85
had low mortality because the army hurried
Nile expedition
preventive health measures.
Ethiopia (Curtin 1998,
(the
We
p. 44)
and out with reasonable
in
use the estimate for the Magdala campaign of 1867-68 for
and the post-campaign, 1882, mortality estimate
Egyptian 1882 campaign mortality was about half the mortality
campaign period,
we use
see Curtin 1998, p. 158). For Sudan,
in the
Egypt
for
immediate post-
the death rate from 1885 at
Suakin (on the coast). This was 88.2 per 1000 mean strength, mostly from typhoid.
very high mortality from disease in Madagascar in 1895
settlers
would have expected because
(Curtin 1998,
p.
In the Mali
Europeans to
p. 89).
This
relative to
this
was not a particularly well-organized campaign
campaign during 1878, the French expeditionary
disease, particularly to yellow fever, in less
is
a reasonable estimate of what
188).
what would have been expected
Africa, precisely because
it
We
in
most
force lost 49 percent of its
than two months (Curtin 1998,
equivalent to an annual rate of 2,940 per 1000
there was an outbreak of yellow fever.
West
is
The
mean
strength. This
is
years, but quite typical of periods
high
when
do not know the true extent of yellow fever
was a major disincentive to
settle
in
and Europeans generally
stayed away from the most dangerous places. Yellow fever was thought less frequent on
the Gold Coast (Ghana) than elsewhere in West Africa, but there were recorded outbreaks
in 1852, 1857, 1859,
and 1862 (Curtin 1998,
p.
65).
In April-May 1873, an outbreak of
yellow fever in the Niger delta killed two-thirds of the Europeans there. This was a quite
typical death rate (see Curtin 1998, Table 4.1, p. 79, for the death rate of
Senegal yellow fever epidemics).
The
case fatality for
was 25-75 percent from malaria and 75 percent
In
adults before quinine
fever killed 279 British soldiers out
was seldom over 120 and often only 40 strong (Curtin 1998,
Assuming a mean strength
in
for yellow fever.
Gambia, from May 1825 to December 1826,
of a force that
nonimmune
Europeans
of 120 gives total deaths per 1,000 of 2,300, which
is
p.
10).
high but
not extraordinary for that time and place. This converts into a mortality rate of 120 per
month, or 1,470 per year.
31
The second column
mortality
number
modern country
to a
if
the country
(1998). In Haut-Senegal (Niger), in 1880-83 there
strength (Curtin 1998,
modern
Appendix Table A2 assigns a
of mortality estimates in
85).
p.
In 1841,
name today
is
the
same
p. 89). In
as in Curtin
was a death rate of 400 per 1000 mean
European mortality on a Niger expedition
Nigeria) was 167 per 1000 per month, equivalent to 2,004 per 1000
per year (Curtin 1998,
settler
mean
(in
strength
a period of several months, 82 percent of the Europeans
died from malaria. At least in this time and place,
it
was almost impossible
for a
person
to go for an entire year without receiving at least one infective mosquito bite.
This second column also assigns the same number to countries that were previously
part of one country. For example, Burkina Fasu, Central African Federation, Chad, French
Congo, and Mauritania were part of French Soudan.
The estimate
for
Vietnam
is
from
Cochin China.
The
column of mortality estimates assigns a mortality number
to a country
if it
neighbors a country for which we have data and has the same disease environment.
On
third
the Gold Coast (approximately southern
Ghana
today) in the period 1824-26, European
annum
troops died from disease at a rate of 668 per 1000 per
(Curtin 1998,
p.
18).
Cote d'lvoire and Togo had the same disease environment and we use the same estimate.
Angola, Cameroon, Rwanda, and Uganda receive the estimate from French Soudan, and
Guinea
is
Pakistan
from the Sierra Leone estimate. The estimate
is
from Bombay.
Surinam
is
for
Morocco
is
from Algeria and
a neighor of French Guinea, for which
The estimate
direct estimate for the period 1819-36.
for
Kong
is
we have
a
from the 1860 China
Field Force.
The
for
third
column
of mortality estimates also assigns a
which we do not have mortality data
who moved
for
number
to several countries
Europeans but we do have data
into this country from a different disease environment, so they did not have
a built-up immunity to local malaria. As Curtin (1968) shows, in
all
situations where
have adequate data, mortality of Africans from a disease environment
comparable to European mortality.
We
1000 (Curtin
We
lower than but
al
1995, p. 491)
and
in
Congo
of 240 per
et al 1995, p. 463).
fourth column of Appendix Table
analysis.
is
we
can have a reasonable lower bound on European
mortality in Kenya of 145 per 1000 (Curtin et
The
for Africans
use the data from the third
A2
provides the data that we use in most of our
column and add
settler mortality estimates for a
few countries that are close by and have the same disease environment as places for which
we have
direct data.
We
use the
New
Zealand estimate
32
for Australia
and the Windwards
and Leewards estimate
for the
Bahamas, Jamaica
for Haiti
and the Dominican Republic
(information from Gutierrez 1986 indicates that these were similar disease environments).
We
also use a
combination of the Gutierrez (1986) and Curtin (1989) numbers to
construct estimates for South America.
aged 40-49
Gutierrez calculates the mortality of bishops
environments
for three disease
10 (low), 11 (medium), and
2.3
in the
Caribbean, Central and South America:
(high) per 1000.
We
assume the
mortality levels in different disease environments was the
mortality.
We
was the same
also
same
as the ratio
assume that the type of disease environment
for bishops
and
We
soldiers.
between bishop
ratio
(low,
between soldier
medium
or high)
then use Gutierrez's ratios with the Curtin
(1998) estimate of mortality of 71 per 1000 for Mexico 1862-63 (a low mortality region
in Gutierrez's classification), to
generate estimates of mortality for Central and South
America.
Note that the estimates we obtain would be virtually the same
(1989) estimate for Jamaica as the
classification).
163.3.
1989).
The
benchmark
The implied mortality
in
(a
if
we used the Curtin
high disease environment by Gutierrez's
Jamaica, using the Gutierrez data, would be
direct estimate for mortality in
Jamaica
is
130 per 1000
Running our regressions using data that takes Jamaica
in
1817-36 (Curtin,
as the base for applying
the Gutierrez's ratios actually strengthens our results.
Gutierrez provides considerable detail on the location of low, medium, and high mortality areas for bishops.
High mortality regions are the Caribbean, low lying parts of Central
This includes Nicaragua, and Panama.
America, and tropical South America.
Medium
mortality locations are Costa Rica, El Salvador, Honduras, Paraguay, and Venezuela.
Low
mortality locations are Bolivia, Brazil, Colombia, Ecuador, Guatemala, Peru, and
Uruguay.^^
Some
of these countries have a large variation in their disease environment,
We
even for a relatively small country.
very unhealthy to Europeans.
high
— the force was there
Mortality
for only
p.
were able to settle at a higher altitude,
favorable.
It
in
the British siege of Cartegena in 1742 was very
two months, but deaths from disease were between 2/3
and 3/4 of the army (Curtin, 1989,
much more
know, for example, that the Colombian coast was
2).
But most Europeans who went to Colombia
e.g., in
Bogota, where the disease environment was
therefore seems reasonable to follow Gutierrez's classification
of disease environments, as the bishops tended to live where European population was
we do not know the precise difference
between the North-East, where much of the early settlement was attempted, and more
southern locations that later attracted population. We have adopted a low mortality estimate as this is
^^It
is
particularly difficult to assign an estimate to Brazil because
in mortality
less favorable to
our hypothesis.
33
concentrated.
The
in Latin
last
column of Appendix Table A2 uses an alternative source to calculate mortality
forces off
West Africa was 54.4 per thousand (Curtin, 1964,
from disease
The
Indies.
British soldiers in
Home
in the
level of mortality
all
locations, but
p. 486).
same period was
at other naval stations over the
Mediteranean, 9.8
in the
West
Average annual mortality, 1825-1845, from disease
America.
in the British naval
Comparable mortality
7.7 in
South America, 9.3
Station, 15.1 in the East Indies,
from disease
it is
for British sailors
and
18.1 in the
was lower than
for
reasonable to assume that the ratio of mortalities
between the various regions was approximately the same.
This change of data does not affect the numbers much.
Chile
move from a mortality
data on naval
with
forces).
little effect
Our
of 71 per 1000 (using the data on bishops) to 68.9 (using the
Table 5 reports regression results using alternative classifications,
on our findings.
baseline estimates in
West African
countries.
We
soldiers.
For example, Argentina and
column 4 imply very high mortality
Some
of these estimates
come from
relatively small
take comfort in the fact that our results are not affected
countries are dropped from the sample.
when
number
is
of
samples of
all
African
Nevertheless, as another check on our results,
constructed a more conservative series of mortality estimates, which
6.
rates in a
we
reported in column
This series uses only Curtin estimates from long sample periods, and assigns those
to neighboring countries with the
same
disease environment.
In particular, this series
assigns the mortality estimate from Sierra Leone to Cote d'lvoire,
Ghana, Nigeria, and
Togo, and assigns the estimate from French Soudan to the French Congo, Kenya, Mali,
Niger, Tanzania, and Zaire.
We
Madagascar (from Curtin, 1998).
an alternative, lower estimate of mortality
also use
Table
shows that
5
for
this different series leads to very
similar results to our baseline series.
There
still
estimates for
remain several significant gaps. Most important, we do not have any reliable
much
of Southern Africa, including Lesotho, Namibia, Swaziland,
Zambia, Malawi, Mozambique, or Botswana.
for Eritrea, the Philippines,
Cape Verde
or
Zimbabwe,
We also do not have any reasonable estimates
Comoros.
Liberia and Thailand were not
colonies and are not included in our study.
Appendix Table A2 reports mortality data
for several countries for
have data on the Political Risk Services' measures of institutions or
basis):
GDP
which we do not
per capita
(PPP
Barbados, Central African Federation, Chad, Mauritania, Myanmar, Mauritius,
Rwanda, and Surinam.
Qualitatively, the
GDP
34
per capita in these countries
is
quite
consistent with the settler mortality
numbers and our regression
results.
Finally, note that these estimates of mortality for soldiers are similar to the
fragmentary evidence available on civilian
European attempts
(Curtin, 1964).
to settle in
settlers.
For example, we know that early
West Africa foundered due
In the "Province of
to high mortality
Freedom" European mortality
from disease
in the first year
46 percent, in Bulama (April 1792-April 1793) there was 61 percent mortality
Europeans, and
in the first year of the Sierra
the European settlers died.
35
more
was
among
Leone Company (1792-1793) 72 percent of
—
in The Effect of Institutions When Other
Endogenous Variables Are Included
Appendix C:Bias
10
To
simplify notation, suppose that
nous,
2,,
exogenous, and another variable that
/?, is
such as prevalence of malaria or ethnolinguistic fragmentation,
=
Yi
is
endoge-
added to the
is
Then, the simultaneous equations model becomes
regression.
where
,
logj/j.
we
interpret
cov
[rji.Si)
presume that a >
and cov
Ri)
(z,,
<
that
0,
0,
<
(p
and
0,
the factor
is
<
tt
on income. Moreover,
as a negative influence
2,
<
We
which implies that
this naturally implies that
be negatively correlated
likely to
2, is
0,
with positive influences on income.
Standard arguments imply that
plima
=a
=a—
~
^
-\
—
k
var{Ri)
where k and Ri are the
coefficient
var{Ri)
and the residual from the auxiliary equation,
=
Ri
Kq
+
+
KZi
Ri,
and so
K
cov{z^,R^)
-—— <
^
0,
var (zj
which
is
negative due to the fact that cov
=
Zi
-
1
We
—
77,,
actually increases
,.
plim
+
(t)T^)
<
0.
4)aRi
•
z^.
<
tt
Now
1,
is
Substituting for k in
£,
(8),
plim
(/>
<
0, ct^^
<
{Oer,
2= a—
0,
(7)
(1
—
+
we can write
^CT^)
^^
(8)
var{Ri)
(p-n)
77.
obtain:
(1
Recall that
(cr^^
the covariance of e and
a^j^ is
Zi is:
+ m).
using this reduced form,
—=a—k
and
we
(t)£i
COv[Zi,£i)
~
a= a—K
the variance of
+
for
so that an increase in the disturbance to the
var{Ri)
where a^
The reduced form
07r
impose the regularity condition
2-equation,
+
-—{{ii
(i?,, Zi)
-
+
07r)
•
and cov{zi,R{) <
control for the endogenous variable
z,,
COv{Zi,Ri)
(l)ol)
L'ar(i?,)
0.
var
[zi)
Therefore, plim
a <
a,
and when we
the coefficient on our institution variable will be
biased downwards.
36
REFERENCES
Acemoglu, Daron (1995) "Reward
pean Economic Review,
Structures and the Allocation of Talent," Euro-
39, 17-38.
Acemoglu, Daron and Thierry Verdier (1998). "Property
and the Allocation of Talent:
A
Rights, Corruption
General Equilibrium Approach." Economic Journal, 108,
1381-1403.
Anderson, Benedict (1983) Imagined Communities,
Verso, London.
Angrist, Joshua and Guido Imbens (1995) "Two-Stage Least Squares Estimates
of Average Causal Response in Models with Variable Treatment Intensity," Journal of the
American
Statistical Association,
Bates, Robert H. (1981) Markets and States
fornia Press, Berkeley
in Tropical Africa, University of Cali-
CA.
Bates, Robert H. (1983) Essays on
Economy
the Political
of Rural Africa,
Cam-
bridge University Press, Cambridge.
Bertocchi, Graziella and Fabio Canova (1996)
Growth? An Empirical Exploration
ment,"
CEPR
Besley,
into the Historical
'Did Colonization Matter for
Causes of Africa's Underdevelop-
Discussion Paper #1444.
Timothy (1995)
"Property Rights and Investment Incentives: Theory and
Evidence from Ghana," Journal of Political Economy, 103, 903-937.
Bloom, David E. and Jeffrey D. Sachs (1998) "Geography, Demography, and
Economic Growth
in Africa,"
Brookings Papers on Economic Activity, 1998:2, 207-295.
Boone, Catherine (1992) Merchant
1930-1985, Cambridge University Press,
Bowman, Larry
W
Capital and the Roots of State
Power
in Senegal,
New York NY.
(1991) Mauritius: democracy and development
in the Indian
Ocean, Boulder, Colo., Westview Press.
Bruce-Chwatt, Leonard
Cain, Philip
J.
J.
(1980) Essential Malariology, London.
and Anthony G. Hopkins (1993)
and Expansion 1688-1914, Longman,
Callaghy,
Perspective,
Thomas M.
York.
(1984) The State-Society Struggle: Zaire
Columbia University
Chabal, Patrick (1986)
New
British Imperialism: Innovation
Press,
Political
m
Comparative
New York NY.
Domination
in Africa,
Cambridge University
Press,
Cambridge.
Chazan, Noami
et al (1988) Poilitcs and Society
Colo..
37
m
Contemporary Africa, Boulder,
Coatsworth, John H. (1978) "Obstacles
to
Economic Growth
in
Nineteenth-Century
Mexico," American Historical Review, 83, 80-100.
Coatsworth, John H. (1982) "The Limits
of Colonial Absolutism: Mexico in the
Eighteenth Century," in Karen Spalding ed. Essays in the Political, Economic and Social
History of Latin America, University of Delaware Press, Newark DL.
Coatsworth, John H. (1999) "Economic and
Institutional Trajectories in Nine-
teenth Century Latin America," in John H. Coatsworth and Alan M. Taylor eds.
America and
the
World Economy since 1800, Harvard University Press, Cambridge MA.
Coplin, William D., Michael K. O'Leary,
to Political
Latin
Tom
Sealy (1991) A Business Guide
Risk for International Decisions, 2nd edition. Political Risk Services, Syracuse,
NY.
Crosby, Alfred (1986) Ecological Imperialism: The Biological Expansion
900-1900, Cambridge University Press,
Crowder, Michael (1968) West
sity Press,
Evanston
of Europe
New York NY.
Africa under Colonial Rule, Northwestern Univer-
IL.
Curtin, Philip D. (1964) The Image of Africa, University of Wisconsin Press, Madison.
Curtin, Philip D. (1968) "Epidemiology and the Slave Trade,"
Political Science
Quarterly, 83:181-216.
Curtin, Philip D. (1989) Death by Migration: Europe's Encounter with
World
in the
Nineteenth Century, Cambridge University Press,
the Tropical
New York NY.
Curtin, Philip D. (1998) Disease and Empire: The Health of European Troops
the Conquest of Africa,
Cambridge University
Press,
New York NY.
Curtin, Philip D., Steven Feierman, Leonard
(1995) African History: From Earliest Times
New
to
in
Thompson, and Jan Vansina
Independence, Longman, London and
York, Second Edition.
Davis, Lance E. and Robert A. Huttenback (1987)
Mammon
and the Pursuit of
Empire: The Political Economy of British Imperialism, 1860-1912, Cambridge University
Press,
Cambridge UK.
Denoon, Donald (1983)
ment
Settler Capitalism:
in the Southern Hemisphere,
The Dynamics of Dependent Develop-
Clarendon Press, Oxford UK.
Diamond, Jared M. (1997) Guns, Germs and
W.W. Norton & Co., New York NY.
Easterly, William
and Ross Levine (1997)
38
Steel:
The Fate of
"Africa's
Human
Growth Tragedy:
Societies,
Policies
and Ethnic Divisions," Quarterly Journal of Economics, 1203-1250.
Engerman, Stanley
Institutions,
Haber
L.
and Kenneth
L. Sokoloff (1997) "Factor Endowments,
and Differential Paths of Growth among
New World
Economies,"
in S.H.
Hoiu Latin America Fell Behind, Stanford University Press, Stanford CA.
ed.
Engerman, Stanley
ing, Suffrage,
L., Elisa
Mariscal and Kenneth L. Sokoloff (1998) "School-
and the Persistence of Inequality
Paper, Department of Economics,
in the
Americas, 1800-1945," Unpublished
UCLA.
Fieldhouse, David K. (1999) The West and the Third World, Blackwell, Oxford
UK.
Preiden, Jeffery (1994) "International Investment and Colonial Control:
A New
Interpretation," International Organization, 48, 559-593.
Gallup, John L.,
Andrew D. Mellinger and
raphy and Economic Development,"
NBER
Jeffrey D. Sachs (1998) "Geog-
Working Paper #6849.
Gallup, John L. and Jeffrey D. Sachs (1998) "The Economic Burden of Malaria,"
Unpublished paper, Harvard Center
for International
Development, October.
Gann, Lewis. H. and Peter Duignan (1962) White
Penguin, Baltimore
Settlers in Tropical Africa,
MD.
Gann, Lewis. H. and Peter Duignan (1979) The
Rulers of Belgian Africa, Prince-
ton University Press, Princeton NJ.
Grier,
Robin
M
(1999) "Colonial Legacies and Economic Growth" Public Choice
Vol. 98 (3-4), 317-35
Gutierrez, Hector (1986) "La Mortalite des Eveques Latino-Americains aux XVIIe
et
XVII
Siecles,"
Annales de Demographic Historique, 29-39.
Gurr, Ted Robert (1997)
"Polity
II;
Political Structures
and Regime Change, 1800-
1986," Unpublished paper, University of Colorado, Boulder.
Gwartney, James and Robert Lawson (1997) Economic Freedom
1997,
The Eraser
Hall,
so
of the
World
Institute.
Robert E. and Charles
Much More Output
I.
Jones (1999) "Why Do Some Countries Produce
per Worker than Others?," Quarterly Journal of Economics, CXIV,
83-116.
Hayek, Frederich A. von (1960) The
Constitution of Liberty University of Chicago
Press.
Herbst, Jeffrey (2000) States and Power
Princeton.
39
in Africa, Princeton University Press,
Holmes,
Kim
Bryan T. Johnson, and Melanie Kirkpatrick (1997) 1997
R.,
Index of Economic Freedom, The Heritage Foundation and The Wall Street Journal, Washington
DC
and
New
York.
Hughes, Robert (1987) The
Fatal Shore, Collins Harvill,
London UK.
Jones, Eric L. (1981) The European Miracle: Environments, Economies and Geopolitics in the
History of Europe and Asia, Cambridge University Press, Cambridge
UK.
Jewsiewicki, B. (1983) "Rural Society and the Belgian Colonial Economy,"
Birmingham and P.M. Martin
New
eds.
The History of Central Africa, Volume
in
D.
Longrnan,
II,
York.
Johnson, Simon, John McMillan and Christopher Woodruff (1999) "Property
Rights and Finance," Unpublished working paper,
Knack, Steven and Philip Keefer (1995)
MIT
and
UC
"Institutions
San Diego.
and Economic Performance:
Cross-Country Tests Using Alternative Measures," Economics and
Landes, David
S.
Rich and Some So Poor,
Politics,
(1998) The Wealth and Poverty of Nations: Why Some Are So
W.W. Norton &
Co.,
New
York.
Lang, James (1975) Conquest and Commerce: Spain and England
Academic
Press,
Shleifer
and Robert
(1998). "Law and Finance," Journal of Political Economy, 106, 1113-55.
La Porta,
W. Vishny
in the Americas,
New York NY.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei
W. Vishny
207-227.
7,
Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer and Robert
(1999) "The Quality of Government," Journal of Law, Economics and Or-
ganization, 15, 222-279.
Lewis,
W. Arthur
(1965)
Politics in
West
Africa,
Oxford University Press,
New
York NY.
Lipset,
Seymour M. (1994) "The
Social Requisites of
Democracy
Revisited,"
Amer-
ican Sociological Review, 59, 1-22.
Lockhart, James and Stuart B. Schwartz (1983) Early Latin America, Cambridge University Press,
New York NY.
Lynch, John (1986) The Spanish American Revolutions, 1808-1826, W.W. Norton
k
Co.
New York NY.
Manning, Patrick (1982)
Slavery, Colonialism,
1640-1980, Cambridge University Press,
and Economic Growth
Dahomey,
New York NY.
Manning, Patrick (1988) Francophone Sub-Saharan
University Press,
m
New York NY.
40
Africa, 1880-1995,
Cambridge
Mauro, Paulo (1995)
"Corruption and Growth," Quarterly Journal of Economics,
CX, 681-712.
Mazingo, Christopher (1999)
"Effects of Property Rights on
Economic
Activity:
Lessons from the Stolypin Land Reform," Unpublished Paper, Department of Economics,
MIT.
McEvedy, Colin and Richard Jones
Facts on File,
New
(1975), Atlas of World Population History,
York.
McNeill, William H. (1976) Plagues and Peoples, Anchor
Montesquieu, Chairles de Secondat [1748] (1989) The
bridge University Press,
&
Co.,
New
Cam-
Spirit of the Laws,
in
Economic
History,
W.W.
Nor-
York.
North, Douglass
der, Disorder
Garden City NJ.
New York NY.
North, Douglas C. (1981) Structure and Change
ton
Press,
C, William Summerhill and Barry Weingast
and Economic Change:
(1998) "Or-
Latin America vs North America," Unpublished
Manuscript, Hoover Institution, Stanford University.
North, D.C. and R.P. Thomas (1973) The Rise
Economic
History,
Cambridge University
Parker, Philip
M.
Press,
A New
Cambridge UK.
(1997) National Cultures of the World: A
Cross-Cultural Statistical Encyclopedia of the World,
port,
of the Western World:
Volume
4,
Statistical Reference,
Greenwood
Press,
West-
CT.
Peemans, Jean-Philippe (1975)
nialism:
The Role
"Capital Accumulation in the
of the State," in Lewis. H.
in Africa 1870-1960,
Volume
4
Gann and
Congo under Colo-
Peter Duignan eds. Colonialism
The Economics of Colonialism, Hoover Institution Press,
Stanford CA.
Reno, William (1995) Corruption and
University Press,
Press,
Staying Poor:
Ghana's
Political
Economy 1950-1990,
Oxford UK.
Roberts Andrew (1976) A History
of Zambia,
Heinemann, London.
Robinson, Ronald E. and John Gallagher (1961) Africa and
Official
Mind
Cambridge
New York NY.
Rimmer, Douglas (1992)
Pergamon
State Politics in Sierra Leone,
the Victorians:
The
of Imperialism, MacMillan, London.
Rodrik, Dani (1999) "Where has
Growth Volume
4,
all
the
Growth Gone?"
Journal of Economic
385-412.
Rogozinski, Jan (1999) A Brief History of the Caribbean, Meridian (Penguin Group),
41
1994.
Turner,
Thomas and Crawford Young
State, University of
Weber, Eugen
ford
(1985) The Rise and Decline of the Zairian
Wisconsin Press, Madison WI.
J.
(1976) Peasants
into
Frenchmen, Stanford University Press, Stan-
CA.
Welsh, Frank (1993) A Borrowed
International,
New
Place:
The History of Hong Kong, Kodansha
York, Tokyo, and London.
Wittfogel, Karl A. (1957) Oriental Despotism:
Power, Yale University Press,
Comparative Study of Total
New Haven CT.
Young, Crawford (1994) The
Yale University Press,
A
African Colonial State in Comparative Perspective,
New Haven CT.
42
Table
1
Descriptive Statistics
Log
GDP per capita (PPPj in
V/hole
Base
World
Sample
1995
8.3
8.05
(1.1)
fl.lj
7
6.5
(1.8)
n.5)
3.6
4
(23)
f2.3)
1.9
2.3
(1-8)
(2.1)
3.6
3.3
(2.4)
(2.4)
1.1
1.6
(2.6)
(3-0;
0.31
0.16
ro.4)
(0.3)
n.a.
4.7
Average Protection Against
Expropriation Risk, 1985-1995
Constraint on Executive in
Constraint on Executive in
1
1
990
900
Constraint on Executive in First
Year of Independence
Democracy
in
1900
European Settlements
Log
in
1
900
Mortality
By
Quartiles of Mortality
rii
iji
r3i
8.9
8.4
7.73
7.9
6.5
5.3
5.1
3.3
3.7
3.4
1.1
4.8
2.4
3.1
5.9
2.8
0.19
0.32
0.26
0.08
0.005
4.3
4.9
6.3
7.2
5.9
2.3
3.4
fl.l)
Number of Observations
64
163
Standard deviations are in parentheses. Mortality
is
14
17
i;
jwtential settler mortaliU', in deaths per
annum per
15
1CM30 people,
sources and methods are described in Appendix Table 2. Quartiles of mortality are for our base sample of 64
observations. These are: (I) less than 65.4; (2) greater than or equal to 65.4 and less than 78.1; (3) great than or equal
to 78.1
and
less than
280; (4) greater than or equal to 280.
Average Protection Against Expropriation
is
measured on a scale from
to 10,
where a higher score
indicates greater
protection against risk of expropriation of investment by government, from Political Risk Services, with 128
observations in the WTiole
World sample and 64 observations
1900, and First Year of Independence
is
measured from
1
in our
to 7,
Base Sample. Constraint on Executive
where a higher score
indicates
more
in 1990,
constitutional
on arbitrary actions by the executive in that year, from the Polity III data set, with 89 observations in the WTiole
World sample and 59 observations in our Base Sample. First Year of Independence is the first year that the countrs'
limits
appears in the Polity
indicates
III
data
Democracy
set.
more democracy, from
the Polity
III
in
1900
is
measured on a scale from
data set, with 89 observations in the
to 10,
where a higher score
Whole World sample and 59
observations in our Base Sample.
European settlements in 1900
from
in the
in
to
1,
the fraction of the population in 1900 that
Whole World sample and 63
1900 are
more
is
was European
or of European descent,
from McEvedy and Jones (1978) and other sources given in Appendix Table A5, with 155 observations
set equal to
1
observations in the Base Sample. Constraint on executive in 1900 and democracy
(the lowest score; for countries that
detailed variable definitions
and sources.
were colonies
in that year.
See Appendix Table
.'M for
Table 2
OLS
Regressions
Whole
World
Whole
World
Base
Base
Whole
World
Base
Sample
Whole
World
Base
Sample
Sample
Sample
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Average Protection Against
0.54
0.52
0.47
0.43
0.47
0.41
Expropriation Risk, 1 985- 1 995
(0.04)
(0.06)
(0.06)
(0.05)
(0.06)
(0.06)
0.90
0.37
1.60
0.92
(0.49)
(0.51)
(0.70)
(0.63)
Dummy
in
for anti-expropriation index
second quartile
Dummy
0.52
(0.25)
for anti-expropriation index
in third quartile
Dummy
0.27
(0.22)
1.50
1.60
(0.21)
(0.26)
2.20
2.00
(0.21)
(0.34)
for anti-expropriation index
in fourth quartile
Latitude
Asia
Dummy
Africa
Dummy
"Other" Continent
Dummy
Adjusted R-Squared
N
Dependent Variable: Log
GDP
-0.62
-0.60
(0.19)
(0.23)
-1.00
-0.90
(0.15)
(0.17)
-0.25
-0.04
(0.20)
(0.32)
0.61
0.59
0.53
0.45
0.62
0.73
110
110
64
64
110
110
per capita (PPP basis)
1999). Average protection against expropriation risk
in
is
1995, current prices, (from the
measured on a scale from
0.56
0.69
64
64
World Bank's World Development
to 10,
Indicators
where a higher score means more protection
against expropriation, averaged over 1985 to 1995, fi-om Political Risk Services. Standard errors are in parentheses. In regressions
including
dummies
for the anti-expropriation index, the
dummy for the first quartile —i.e., with highest risk of expropriation— is the
dummy for America is omitted. See Appendix Table Al for more detailed
omitted category. In regressions with continent dummies, the
variable definitions and sources.
. '
r
>
m
c
r
.0.
>
r-
S
,0-
(JQ
a>
r
tn
c
^
(JQ
r-f-*
:5
2
c
09
D.
^
O
<T
H
ro
TO
C
3
D-
Q-
3
3
rt
3
CD
Q.
c
fD
~-i
3
n
ft
D.
!-*
m
_
TO
.
O
Q
£?
pj
1W
re
r»
00
:r.
c/5
^
3
_,
§
re
3'
a.
3
3 m
a. X
m ft
3 c
<
m
3
3'
3
^^
m
<
n>
3'
'
31
—
B
3
—
o
S
TO
B
BJ
Q,
n i
a.
5'
S
C/l
-a
0
O
o
(t)
<
>-"
^—
VO
3
in
—
3
-a
5'
o
(D
3
w
CO
I
CO
XI
00
XI
^
3
3.
C/2
O
r-
T3
2
a>
CO
^
a
-<
^-
o
P
w
"
4^
is
s
O
Pi
u
o o S
OJ
OJ
^
b
p>§
p
to
o
^o 00 a
—: to
4^
~^
^
O
00
s S
o
s
5
3
-a
0^
b
s s 4^
o o
re
s
re
Qc:
§ 3 3
=
m
3.
m
"^
fD
2
w'
fD
3
ft)
-^
P
5'
1
-f^
°" I
° s
§
o|.
—
00 2
i
^
5'
ci
OS
-
°
^
<
:
.
S.
S^
tra
Dre
„
^
S
2
-•
&
w ;
re
^'
TO
=i
re
"^
-
I
,
-
3
3
^
S,
6
3
9-
2-
"^
i'
I
i «
i
[a
^
""
5' EL
O w CA
3 re _.
X 3
!2.
O
!^ T3
H. 3 c"
os -a
-!
re
j^
„
O
o
2 =
3 o
O
re
-'I
-T-t
a-
re
^
2:
o
a
o
1^
to
CO
i^
ON
10
'-^
Oo
V'
P 6
I—
o
P 6
to
On
<=>
re
a
a
re
g
w
3-
Co
u5
Ire
s
re
a
:i
re
s
"
n re
o <
-I
ja
o'
I
a
o
5' v:
:r.
to
u>
re
o
io
5'
\-l
2 S
-J
to
H
to
m
C
o
00
4^
to
n S
3
5
° ° T
3
3
re
re
to
o
O ^
to
bJ
"=
re
a
re
o
00
4^
^
\_:
^
re
to
a
00
8 ^
5-
™
^'
to
o
00 o
s
o
re
»
o
<=>
o ?°
o o
0\
2
S'
re
S
o
0—0
-"
s S o
>
°
"«
2^
to
4^
C3-
Ess
»•
o
0\
°°
,
Table 4
IV Regressions of log
GDP per capita
Base
Base
Sample
Sample
Base
Base
Base
Base
Sample
Sample
with
with
without
without
Sample
Sample
Base
Base
neo-
neo-
without
without
Continent Continent
Sample
Sample
Europes
Europes
Africa
Africa
Dummies Dummies
(1)
(2)
(3)
(4)
(5)
(6)
(V)
(8)
Panel A: Two Stage Least Squares
Average Protection Against
Expropriation Risk 1985-1995
0.94
1.00
1.28
1.2
(0.16)
(0.22)
(0.36)
(0.35)
Latitude
Asia
0.58
0.58
0.90
1.00
(0.10)
(0.12)
(0.23)
(0.4)
-0.65
0.94
0.04
-0.94
(1.30)
(1.50)
(0.84)
(1.5)
Dummy
Dummy
Africa
"Other" Continent
Dummy
-0.86
-0.96
(0.35)
(0.43)
-0.50
-0.48
(0.32)
(0.36)
-0.62
-0.65
(0.6)
(0.6)
"
B: First-Stage for Average Protection against Expropriation Risk
Log
Mortality
-0.61
-0.51
-0.39
-0.4
(0.13)
(0.14)
(0.13)
(0.14)
Latitude
Asia
-1.2
(0.22)
-0.48
-0.4
(0.24)
(0.16)
(0.17)
2.00
-0.11
1.00
2
(1.30)
(1.50)
(1.40)
(1.40)
Dummy
"Other" Continent
1985-95
-1.1
Dummy
Africa
in
0.3
0.4
(0.5)
(0.5)
-0.26
-0.25
(0.41)
(0.41)
1
-0.87
(0.68)
(0.68)
0.26
0.27
Dummy
Adjusted R-Squared
0.26
0.27
Average Protection Against
0.52
0.47
(0.06)
(0.06)
64
64
0.12
0.1
0.44
0.45
Panel C: Ordinaiy Least Squares
Expropriation Risk 1985-1995
Number of Observations
"Average Protection Against Expropriation Risk 1985-95"
against risk of expropriation of investment
A
reports the
two stage
by
0.49
is
(0.08)
(0.07)
60
60
column
(full results
is
omitted.
(0.07)
37
measured on a scale from
with log
GDP per capita
protection against expropriation risk using log settler mortaliry; Panel
America
0.48
GDP per capita
B
(PPP
0.47
0.43
0.41
(0.08)
(0.06)
(0.06)
37
64
64
means more protection
Appendix Table Al for more detail). Panel
where
to 10,
the government, from Political Risk Services (see
least squares estimates
coefficient from regressing log
0.47
a higher score
basis) in 1995 as the dependent variable and instrumenting for
reports the corresponding
on protection against expropriation
risk,
first
stage.
Panel
C
reports the
with the other control variables indicated
not reported to save space). Standard errors are in parentheses. In regressions with continent
dummies,
the
OLS
in that
dummy
for
^
s
B'
>
< n 3
o o
3
E"
2. H"
c in
c VI
3 3 3
3
3
n
o
X
H
,—
'-^
^-^
Op
„—
,—
-J
Q-
CL
r;
lO
5
<T)
rS
3
3
W
3
CL
m
ni
o" 3^
3
03
>
o
s
k
o.
o
x' 3_
H
03
ni
o
> 3O
c
:3
13.
cu
M_
-^
p
o o 5' 3
Cl
cn 3_
c -^
3
o
r-
O
3
3
o
3 3
o 00
ja
>
oo'
3
a>
zr
o'
0=
•5"
£'
>^
T
OQ
c'
o
m >
r
o
>
tn
-o
i^
3
S
P3
3
D3
<
5'
=
D3
cr
3
O-
o
3
n
- >
VO
oo
(TO
w
o
o o
o o
T3
o
5'
o _
—
3
—
OJ
P
^—
j:^
P3
=^ 3.
m
— p
oo ^J
o
T3
-a
3
::.
B.
vO
-a
o'
to
5'
w'
D3
-t3
re
3* S
5'
-'
3
o
pa
?3
O
D
o
-t
(TO
5'
?o
Jlo"
o
"I
T3
O
o
o*
OQ
3
a>
&>
S'
a
~0 s
m
00
0^
cu
3.
<
P
'
a o ^
^
<7
i"
3
o
o
3
—
S
oo
m
~a
a.
o 3
o" m
o 3*
3 ,—
-b
o
o C ~^ 3 ^O
o
oo
b!
O
Oi
<' a;
ni
o'
O
00
>
-o
D
n
^
3" CL
o
3
^-v DQ
3 3 3
in
t/3
3
OQ
3
fjQ
c
3
NJ
oo
n
c
m 3 3
o
CO
n
o
r?
3._
>
fD
EB
3
P
3
'
.
^
>
ao
^-
C3
-i-
<
-
73
C3
3
o
5
CL
3 u
03
?
o o
<
o
^
>
O
t/j'
5
M
-a
o
"0
ON
o
;
O
K)
k)
OO
o o
;__.
o\
^>
a
3 3
2.
3
^ p
f^
o o
a
to
ON
^§
^^
oo
Q
't
o
3
tjj
2:-§
(-n
M3
to
00
rt
1-^
TO
Q
03
:i;.
CL
!^
3
nr
ni
03
fT*
^
~
f^
3
C/3
<
t/3
^
Ci3
on
2 "O>
3
>
m:i
3
3
s
>^
3
^-^
^
IiT'
—
o o
to ^
o to
S
3-
O.
>
3
.^
o
3
T3
<=
ru
B.
5'
f?
— 3
3 05
o'
>
5'
3
>
o'
D3
o
p
Q.
ta
O
<'
2:
m
3
(-)
73 (^
a.
5?
o-
c
<
3 n
c Q
O
=71
„^
3
o
S
n
< >
c
B-
Q
^ —
o
3
o
=^
^
>
Q. i;
o
"-h
3
3 2
m yi (TO
5
c
D
o
o
3
-O
"
o p ;5_
5
o ?
°
3
r,-
o
(TO
_
3.
^ >
3-
m o
— o
4^
> ^
K>
^
&3
£i
TO
ro
a
o
O
o"
c >
3 _
S &
m
C
n
^
o O
n 3
D3
o —
to ^
3 o
•5
73
c
h:
(It
C/1
g 3
*Co
3- ^-
O 3
a
a
o o
i^
to
fD
3
3
3
o
OQ
Co
(X)
o
O
3
era
^< 3
>
ni
3
<
^^ a\
—
Oh
n
3
:3
TO
y
1
1
5'
2
5
3
a
H
(TO
"
&3
03
Cu
3-
ra
eg
re
Cl^
3' =- 2.
3
oo
00
ro D.
D5
cn
O
CL
Table 6
GDP
IV Regressions of log
per capita with Additional Controls
Base
Base
Base
Base
Base
Base
Base
Sample
Sample
Sample
Sample
Sample
Sample
Sample
(2)
(3)
(4)
(5)
(6)
(7)
(1)
Panel A: Two Stage Least Squares
Average Protection Against
1.10
Expropriation Risk, 1985-1995
1.20
0.29
(0.19)
Latitude
Dummy
Belgian Colonial
French Colonial
Dummy
Spanish Colonial
Dummy
French legal origin
1.1
1.2
0.92
1
1.10
(0.19)
(0.29)
(0.15)
(0.25)
(0.29)
-L2
-1.1
-0.94
-1.7
(1.50)
(1.60)
(1.50)
(1.6)
2.4
2.6
1.7
(1-20)
(1.40)
(1.50)
0.66
0.76
0.27
(0.38)
(0.47)
(0.78)
1.1
1.1
0.07
(0.33)
(0.39)
(0.76)
dummy
0.89
0.96
(0.33)
(0.39)
0.25
(0.81)
p-value for Religion Variables
[0.004]
[0.001]
Panel B: First-Stage for Average Protection against Expropriation Risk
Log Mortality
-0.57
-0.48
-0.54
-0.44
-0.58
-0.44
-0.48
(0.16)
(0.13)
(0.14)
(0.13)
(0.15)
(0.18)
1.8
(1.4)
French legal origin
Dummy
Dummy
French Colonial
Dummy
1985-95
(0.14)
Latitude
Belgian Colonial
in
[0.41]
2.1
2.5
2.3
(1.3)
(1.5)
(1.6)
-0.67
-0.69
-0.1
(0.32)
(0.32)
(1.1)
-2.9
-2.6
-2.6
(1.3)
(1.3)
(1.7)
-0.51
-0.6
-0.2
(0.43)
(0.43)
(1.00)
-0.66
-0.62
0.15
(0.40)
(0.40)
(1.00)
Adjusted R-Squared
0.31
0.31
Expropriation Risk 1985-1995
0.54
0.5
0.56
(0.06)
(0.06)
(0.06)
Spanish Colonial
0.3
0.28
0.3
0.3
0.56
0.53
0.47
0.47
(0.06)
(0.06)
(0.06)
(0.06)
0.31
Panel C: Ordinaiy Least Squares
Panel
A
reports the
Panel
B
reports the corresponding
significant in the
two stage
second or
least squares estimates
first
first
stage.
with log
The Dutch,
GDP per capita (PPP
Italian,
(full results
colonial
dependent variable, and
dummies
are not
stage and are not reported (to save space); the religion variables are included in the
stage of columns (5) and (6) but not reported here (to save space). Panel
GDP per capita on
basis) in 1995 as
German and Portuguese
average protection against expropriation
risk,
C
reports the
OLS
coefficient
with the other control variables indicated in that column
not reported to save space). Standard errors are in parentheses. All regressions have 64 observations.
religion variables are 100 times percentage of population that are Catholics,
base case. Our sample
is all
either
first
from regressing log
The
Muslims, and "other" religions; Protestant
French or British legal origin (as defined by La Porta
et al
1999.)
is
the
-
a.
.
m
X
3'
O. o_
3
-
i
3
3
3' OQ
c
3
3
»
3
P_
"1
5'
O 3 S
3.
g^
52.
3
3
3
5
o
S.
3-
-,
ni
-,
-a
-a
c
a-
o
3- 3-
^ P
3 r
C/l
O
3
^
>o
o
xs
a
p
3
C
P
5
^-.
m
c
00
?3
p
^
p'
o
?3
c/i
^-'
CD
fi>
&)
?<
o
I
O.
T3
n
CI—
3
5-
w
fO
-1
crq
a>
Q.
4^
^
^§1
^
ON
o
n
I
3 £
-
CI.
f?
£
<
ft
O
r-f
CO
3
<
3
—
-a
a.
3
=
S O
3 S
(ra
n
"T
"°
era
S
,.
--
c/)
3
—
S
o
:g
era
O
O
Id
<T>
o
to
S:
3
in
a^
=^
o
oa
O
o
-0
p"
a(D
(T)
5'
n>
P
3'
3'
ft
B.
B
1—
5?
55' 0'
?r 3
>
^
Oq
00 P
n
.
3'
<
p
1
^
VO
p'
a;
w
r-f
L-h
ft
'^r,
£^
3
P
to
hJ
t>J
to
to
O
<i-,
In
C/5
o o
--.
to
to NO
00 \D
'
S
rf
Tt
CO
^
o P
O -^
T3
P
C"
O
,^
5
o o
re
i^
ON
o
00
P
—
to
"a
"a_
ft)
3
o
S,
i>J
— o
j^ o
CO
p
!:£
'5 3
^
CD
p
on
"H-
(D
O
o P
Sr
— -aX
o o
o
to
-J
1
."H
^
00
(^
to as
p
3
g
o
o ^
3
en
P
O
I
OQ
re
o
c;
cp io
00
J^
^
i=.
O
o
1^
o o
LO
O to
O
a
C/5
OS
--4
a.
3
p
Cp
e
3
3
w
(T)
C/3
-t-
o P
On ^
3-
O
O
o o
to
io NO
00 ON
'a
re
Co
--J
3
Fe
ft)
3
W
3
3 <
< 3
3
:2. X
"^
ii
-r
OQ
O
O
00
o
^ p 00
P P
ON
NO
O NO
05
^
3 p
L/j
a\
S- 3' S"
<
P
"
3
R
3
^s-
?;
P
(JQ
o
S
c
"
w
On
-^
,
^
,
O O o
;f^
to
K^
to
'-f^
.p^
o P
o P O
W P 4^ ^
K
C3
o o
i
to
CO
^
to
--J
Lh
NO
-o
o
p
re
-
n
CO
re
3
OQ
o"
00
re
-1
<
^
m
w
ro
—
ao=. 03
o
oc
3
a>
2.0.2:
3.
CO
^
£?
3
c
3
CO
o o
OQ
ft
r^
3
3
TS
m
p
t^
1>J
a.
0'
7;^
ro
re
fS
3 0? S
9:^ 33'
X »
IZl
•-t
to Co
o P
o J^
ON
CD
3
" n
>
— 3o
-K o
o* 3
3
I:
3 =:.
o
c
3
^
n>
^-i
S.
ft
o. g.
3-
:i:
4i-
-a
T^
C:3
—
5-'
,
4^
TO
-^
<
"(&!::
?r :3 —
r
C/D
00
fn
3 c
= 3
5
= §
^
f>
^
o
=i
fo
I
o>
—
<
p
n
o
3
—
~
o
3
S
n>
r-f
1
o"
3-
p
c
"-S
^3
,
P
C
51
a^
00
'
r m >
Ei
a <
c
p
D- ^ OQ
m
^
1' ^
^
^3
o
^.
C
o
n >
r
>
Q.
c
O-
f
t
p
NO
n
p
o.
a
•-'
p.
:f'
^
:=i'
» O
<
s ^ «
—
ra
C
5'
—t
a>
^
,
.
g.
51
?r
re
-o
;;
-t
3
r?"
ao
»;
&
fB
r-K
c"
en
ft)
3-
-"
^
s,
i
o o
era
^
g
X)
1^
EJ
S.
S-
-
1:
H?.
CO
S -
^
3
--
"S
a
9
^
5'
- 3-
2
5'
c
O
TO
-t
n>
O
fa
T3
TO
o
o
m
m
en
O
C
o
ft
TO
S-'
^
i'
-'
ft
25
§..
>
^
TO
OO
3
^
p}
3.
3
en
1
.
,
o
3
ft
ON
oo
o o
O
:^
ft
fa
fa
£.
a-
"-rt
t
C
GO
o_
a"
o
t_»s
T3
C
I
" 3- S S p
P=
C — ,—
;3
—
9 ^
S <
-•
3
'^
t"
"
^
o ro
n
-J ~ 3 n
—
XI f-^
3 Tj
^
o
—
w o :: ^
&J
-V s
cr. rt
a" n o oo
3
S
^
a o -, _.
a. 5^ -J ^
n>, v:
< ri ft 3
s-' a n> crq
D. p o ^ f^ ^O
§
S I- S
5"
TO
ft
>
t-d
S^.
I
oo
(^
R
o
O
^ B ^
C
a.
^1^
S
73
v:
XI
§
3
» I
_
_ I
I
'^
5'
g
3
Hi
I
3
<
1. 1; ::
o
a.
p
3
&3
CO
g
h^
ft
Q.
1
< <
TO
D.
r-^
r
r
o
r
&
>
CO
£^
3
ta
to
3-
^
^
O P
M
^o
^
2
«
o —
!fi
to
U)
o o
O^
TO
p;;
<-"
CI
I--
o
oo
so
s
^'
5'
ft
'^
c
o
ft
^
n
>-^
^
<^
3
c/3
p
VO
^
ft
o
cr
c
3
ft
>
a
X
o
•§
4^
"^
-^
i?
fa
si
ft
ff
D- E. a,
CO
to
-^
"!-
1
o
^ o g
3 c P O
-
OS
-^ S
O
CO
o o
w
^^
oo K)
2
ft
3
o
?r
"2-
en
rt
!:?>
O
o
"2
a ift 3EL ^o ^ ft
« :r. ft 2 o «
l^
,-, C3
M 3 O
o c ^ P ^
"
4^
00
4^
--S
o P
P
to \0
,
oo
e-fi
SO
Co
S
Oq
^^
oo
<
fa
4.
TO
3
?3
"H.
ft
ft
TO
TO
-< P
g- :^
g-
00
O-i
ft
--
s
r-^
P
- ^Co
lo
oo
s
;_^
VO
3
-a_
o o 3 o
SJ ft ^ S-.
O
CL "^ 3 rn
ft
ft
ft
W M
O " p
^
o
ft
-1
p
<
3 S 2. — 00
S^ '::^ '^ 3 ft
fT o 2- ^ -g
C o ft ^
p^ fe '^ oj o
c
a.
OO
5-
P
§
3
i=r.
5
ra
i-^
.
,
S g.
^ §
D. w T w
<
ft
o
-r
Q. cr rt
-
<
Q
ft
2
o
ft
P
ni
ft
ft
-
to-
fa
on
ft
^
SO
2.
ft
-t
CO
fa
ft
"•
S--
o:
^
OO
.
o
ft
^
ii
2.
(TO
•
OO
S
CO
;n
fa
^
ft
ft
O
=>
—
SO
^
p § S " o
K -a 2. S „ o
O 2 3- rt
ft
pj
"> S
»
2 TO
^
—
3
5 - ^
o
o ft
X C
2
o
o
-o
p
3
<
Q.
a3^
CO
ft"2rtX3-r;oN
^
03
1-n
s^
so
so
g
"
o"
TO
00
— o O
H —
:s.
""
"
=
S o
02. S
—
^
=1
9- OS TT -Q
oo r^
Q < 3
-a
O
3
en
O
f»
wi
to
so
OS
o
-^
^ p
~0
(sj
o o
so
-^
'^r\
to
^
o o
•""
171
to
oo
bo
so
CO
fa
-^ 3
DO
(a
en
ft
o
fa
T3
H
fa
o;
ft
oo
—
=
,
»==:."
S_
o
ft
a H
f
o
5
(TO
o 2
c
m
>
X <
00
-a
3
o
oo
-I
3
^
c
C
p
o S 3
3
_fr
2
-TJ
3'
O
3
P
00
Lrt
a>
c
3
P
_ >
VO
CTQ
(TQ
>:?
—
tra
O
w
g
<
re
p
-t
o
5'
o.
•<
re
p
-1
tra
£
5'
o
a.
m
c
O
o
n
o
3
3
P
5"
§.
3'
o
3
o
3
3
3
c
a.
P
3
Q.
re
3
o
re
<
re
P
-0
70
S
to
?r
o
3
^-A
>
rt
a
re
§
s
VO OO
OO
£.
5'
re
3'
1
o
o
o
o
a.
CD
•T3
o>
re
O
-a
C/l
re
o
c
m >
X!
"O
-1
3
M3
-O
r
re
•n
re
•S
vo
I^
w
'~*'
3
•
Q.
cr
o
:ii
-t
t/3
^ U " <
f?
3
o
o
-1
p
o
:ri
$
<
cr
a
re
O
3
~
:;
CO
-^
P
t«
3
>
O)
re
3
='
•
,
O
Q.
m
t
tro'
o o
o
§3
m
"
q
n TO <
o 3 n
S"
3. °
3
Q.
s.
k)
O
^
bo
00
o
>_.
3 p
o
-a
5
w
—
!
P o
to
LO
O
D
OO
T3
— o
^ ~
a k> On
s o
"0
TO
O o
t^
^ ^
S
03
>—
a
n
to
3 Q
DO
n>
o
00
o
C;
3
to
-1^
Q.
re
re
a-
re
ci.
E3
c
a
s
w
p
Co
TO
o
o
o o
oo
OJ
oo
-C^
p
_^
.a
_
>3
^ X
f~->
oo
§
Cd
"^
re
re
_
re
D.
C/l
C3
5
—
,
o
3
c^.
^ p
o o
to Ln
w to
>5
3
^
3
op
,
—
O
;?
T3
—
L^
o o
D- 3" w'
c
Oi
Q.
5" w'
tn
UQ
:j-
^
ro
w
C p
5
cr
re
>'
ON
r?
s.
<*
3
O
3
o
^
c;
3
re
3
no
5'
Q.
p
Ji
2.
—
a Q.
C
w
ao TO5'
re
re
3
re
=r
CO
c
a.
re
re
^ 3
2
C/3
re
S-
5
re
o-
re
{yi
3
a
<
^5
o —
to
L*J
»^
s::;
" o
§ 5 !^
op
o p
oo
to
^^ —
o;
sa
'
-.
o —
oo OO
o ^
a
§
^
o'
S
3
03
T3
03
3
P
re
o o
^ k)
:-'
Li
o o
:—
o
oo
o
^
--J
00
?=
^ S
§ 2
5'
P
CD
a
re
re
f^
_^
re
t
X^
p
O
CO
p
in
re
to
D-
;^
'^
3re
re
3
o
o
o
-t
re
o
2
CO
o
p
3
-a
o
X
^
TO
re
C
p_
r=
g;
TO
3
o
j^
CO
'
<
P3
to
CO
<
O
s
to
oo
O
3 p
re
p
re
EL
re
a
i^
P
3
_.
o
S:
(^
i^-
TJ
re
C/3
o —
ro
a_
K
o
S
4^
o
CO
o —
to
o
O
3- 5"
re 00
27
o
o
^ —
to
—
— o
t^ to
u)
o
o
'—
OS
LO
<-fi
•^
s
C/5
P
03
-i
re
re
o
o
T
3
3q
3 H
3
^1
S?.
3
C
3
o
3
5
tfl
—
S
S
o 3
" "
S
^
9
3'
3
q
3
."
3-i-
S ?
3
i;;
" i
f
o
C/1
-
i
5
>
'
"3
o_
.5
I
c/3
-a
I
°
2.
C/5
T3
O
^
O O
£.5
i
3
^ 5
-
rt
^
c
=
•3
-
^
9-
t^
—
o^
*
S
a-
;
<'^ < ^
<
? i s I
p 3 U 3
Z o Z 5"
"
c
g-
w O
i
3 3.=-
g^
O
3
-~
3
~
a
-!
O
7
O
T
O
5
7
C
2 5" ?
3 - s"
i" s
Sw
-3
re
2_
m
o "
in
•3
-3
n
n
3
—
Appendix Tab eA2
Data on Morta Ity
Fourth
Log
GDP
Average
mortality
Protection
estimate
Alternative
Abbreviated per capita
Against
First
Second
Third
(used
name used
Expropriation
mortality
mortality
mortality
main
mortality
Curtin data
Risk 1985-95
estimate
estimate
estimate
analysis)
estimate
on
in
graphs
(PPP)
1995
in
version of
Fifth
in
Africa
Former Colonies
Argentina
AGO
ARG
Australia
AUS
9.90
9.32
Burkina Faso
BFA
6.85
4.45
Bangladesh
BGD
6.88
5.14
Bahamas
BHS
BOL
BRA
BRB
CAF
CAN
CHL
9.29
7.93
8.73
Angola
Bolivia
Brazil
Barbados
Central African Fed.
Canada
Chile
7.77
5.36
9.13
6.39
280
280
280
280
68.9
71
68.9
8.55
8.55
8.55
280
280
280
280
280
71.41
71.41
71.41
71.41
71.41
7.50
85
85
85
5.64
71
71
71
7.91
71
71
71
85
280
85
280
85
280
71.41
85
9.27
7.19
9.99
9.73
9.34
7.82
16.1
85
280
280
16.1
16.1
85
16.1
16.1
16.1
68.9
71
68.9
668
280
240
668
280
240
483
280
280
CIV
7.44
7.00
Cameroon
Congo (French)
CMR
COG
7.50
6.45
7.42
4.68
Colombia
Costa Rica
Dominican Re
COL
8.81
7.32
71
71
71
CRI
8.79
7.05
78.1
78.1
78.1
DOM
8.36
6.18
Algeria
8.39
6.50
8.47
6.55
7.95
Ethiopia
DZA
ECU
EGY
ETH
Ghana
GHA
Guinea
Gambia
Cote
d'lvoire
240
668
280
240
130
130
130
130
78.2
78.2
78.2
78.2
78.2
71
71
71
6.77
67.8
67.8
67.8
67.8
67.8
6,11
5.73
26
7.37
6.27
GIN
7.49
6.55
7.27
8.27
26
668
483
1470
26
668
483
1470
26
483
483
CMS
GTM
GUY
26
668
483
1470
8.29
5.14
71
71
71
7.90
5.89
32.18
32.18
32.18
HKG
HND
10.05
8.14
Honduras
7.69
5.32
Haiti
HTI
7.15
3.73
India
DNI
IND
7.33
8.27
48.63
7.33
7.59
Jamaica
Kenya
Sri Lanka
Morocco
Madagascar
Mexico
JAM
KEN
LKA
8.19
7.09
7.06
6.05
7.73
6.05
MAR
MDG
MEX
8.04
7.09
6.84
4.45
8.94
Mali
MLI
Malta
MLT
Myanmar
MMR
Mauritania
Nigeria
MRT
MUS
MYS
NER
NGA
Nicaragua
New
Ecuador
Egypt
Guatemala
Guyana
Hong Kong
Indonesia
Mauritius
Malaysia
78.2
1470
32.18
32.18
32.18
14.9
164.66
14.9
14.9
14.9
78.1
78.1
78.1
130
130
48.63
170
48.63
170
130
130
48.63
48.63
170
130
48.63
170
130
145
170
130
280
69.8
69.8
69.8
69.8
69.8
69.8
78.2
78.2
78.2
78.2
536.04
536.04
536.04
536.q4j
302
7.50
71
71
71
71
71
6.57
4.00
2940
2940
2940
2940
280
9.43
7.23
16.3
16.3
16.3
16.3
16.3
16.3
5.77
34.6
34.6
34.6
34.6
34.6
34.6
280
280
280
280
280
30.5
30.5
30.5
30.5
30.5
30.5
17.7
17.7
17.7
17.7
17.7
17.7
400
400
2004
400
2004
400
2004
280
483
163.3
163.3
163.3
8.55
8.55
8.55
8.55
8.55
36.99
36.99
36.99
36.99
163.3
7.41
9.05
170
130
145
130
145
8.89
7.95
6.73
5.00
6.81
5.55
NIC
7.54
5.23
NZL
9.76
9.73
PAK
PAN
PER
PRY
7.35
6.05
8.84
5.91
163.3
163.3
8.40
5.77
71
71
71
8.21
6.95
78.1
78.1
78.1
Rwanda
RWA
6.48
Sudan
SDN
SEN
7.31
4.00
7:40
6.00
Niger
Zealand
Pakistan
Panama
Peru
Paraguay
Senegal
8.55
164.66
280
280
280
280
88.2
88.2
88.2
88.2
36.4
164 66
164.66
164.66
164.66
164.66
Fourth
Log
El
Salvador
Second
Third
(used
name used
Expropriation
mortality
mortality
mortality
main
mortality
Curlin data
Risk 1985-95
estimate
estimate
estimate
analysis)
estimate
on
graphs
Tunisia
Tanzania
Uganda
Surinam
Chad
Togo
Trinidad and
Tobago
Uruguary
USA
Venezuela
Vietnam
South Africa
Zaire
Alternative
First
SGP
SUR
TCP
TGO
TTO
TUN
Singapore
estimate
Against
(PPP)in
1995
SLE
SLV
Leone
mortality
Protection
Abbreviated per capita
in
Sierra
GDP
Average
6.25
5.82
7.95
5.00
10.15
9.32
483
17.7
483
17.7
4.68
78.1
78.1
17.7
17.7
17.7
32.18
32.12
280
668
85
63
280
668
85
63
145
280
280
668
85
63
280
483
85
145
280
280
280
71
71
71
15
15
15
78.1
78.1
78.1
85
63
7.45
8.48
6.45
TZA
6.25
6.64
145
UGA
URY
6.97
4.45
280
9.03
7.00
USA
VEN
10.22
10.00
9.07
7.14
VNM
7.28
6.41
ZAF
8.89
6.86
ZAR
6.87
3.50
15.5
483
78.1
32.18
8.77
15
483
17.7
6.91
85
Africa
483
32.12
280
6.84
7.22
483
version of
Fifth
in
15
15
61
140
140
140
140
140
15.5
15.5
15.5
15.5
15.5
240
240
240
280
European Colonizers
France
Britain
A
blank indicates missing data.
When
20.17
20.17
20.17
20.17
20.17
20.17
15.3
15.3
15.3
15.3
15.3
15.3
mortality varies across different cities for a country,
we use
the lowest rate.
Countries
have reasonable mortality estimates, but they cannot be included in our basic sample as other data are missing.
shows that the ratio of mortality for bishops aged 40-49 is, by disease environment: low mortality 10 per 1000;
medium mortality, 11 per 1000; high mortality, 23 per 1000, We use these proportions and the death rate in Mexico from
Curtin to infer mortality in Latin America. The results would be essentially the same if we used Curtin (1989)'s Jamaica
estimate as the base case for calculating Latin American settler mortality using the Gutierrez estimates.
in italics
Gutierrez
Curtin (1964) reports average annual mortality from disease
in the anti-slavery blockade (West Africa) as 5.44%, and
South American Naval Stations as 0.77%. Mortality at other naval stations is consistent with
the Curtin and Gutierrez estimates (comparing the ratios with West Africa): Mediterranean (0.93%), East Indian (1.51%),
mortality from disease in the
West
The
Indian (1.81%),
Home
Station (0.98%).
alternative mortality estimates are calculated as follows.
First mortality
estimate: data from Curtin (1989), "Death by
Second mortality estimate: data from "Death by Migration" plus earliest data for each country from Curlin
(1998), "Disease and Empire". Third mortality estimate: including neighbours with same disease environments and African
mortality for Kenya and Congo, from other Curtin sources discussed in the appendix. Fourth mortality estimate: using
Gutierrez (1986) data on bishops to estimate mortality in South America and the full set of Curtin estimates of mortality. Fifth
mortality estimate: using Curtin (1964) data from naval stations for Chile and Argentina. Alternative version of Curtin data:
using long sample periods from Curtin (1989) and (1998) and data on bishops to estimate mortality in South America.
Migration" only.
Appendix Table A3
IV Regressions using
GDP per capita
Base
Base
Sample
Sample
in
1975
Base
Base
Base
Base
Sample
Sample
with
with
without
without
Sample
Sample
Base
Base
neo-
neo-
without
without
Continent Continent
Sample
Sample
Europes
Europes
Africa
Africa
Dummies Dummies
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Panel A: Two Stage Least Squares
Constraint on executive in 1970
0.54
0.6
0.64
0.7
0.31
0.28
0.48
0.58
(0.15)
(0.23)
(0.31)
(0.35)
(0.13)
(0.17)
(0.18)
(0.3)
-LI
-0.8
0.5
-1.4
(1.90)
(2.10)
(1.50)
(1.9)
Latitude
Asia
Dummy
Africa
Dummy
"Other" Continent
Dummy
-1.1
-1.2
(0.48)
(0.60)
-0.65
-0.64
(0,34)
(0.40)
-0.9
(0.9)
(0.7)
Panel B: First-Stage for Constraint on Executive
Log Mortality
1970
-0.82
-0.66
-0.54
-0.5
-1.5
-1.2
-0.79
-0.62
(0.21)
(0.23)
(0.23)
(0.24)
(0.44)
(0.47)
(0.27)
(0.29)
3.50
1.30
3.30
3.1
(2.20)
(2.60)
(2.80)
(2.3)
Latitude
Asia
in
Dummy
Africa
Dummy
"Other" Continent
Dummy
Adjusted R-Squared
-0.4
-0.12
(0.9)
(0.95)
0.1
0.07
(0.70)
(0.70)
1.7
1.6
(1.10)
(1.10)
0.2
0.21
0.2
0.22
0.16
0.12
0.08
0.07
0.13
0.1
0.11
0.09
(0.05)
(0.05)
(0.05)
(0.05)
(0.06)
(0.06)
(0.04)
(0.04)
59
59
55
55
3!
31
59
59
0.07
0.06
0.26
0.25
Panel C: Ordinary Least Squares
Constraint on executive in
1
970
Number of Observations
Dependent Variable
to 7,
where
is
log
GDP per capita (PPP basis)
a higher score represents
estimates and Panel
B
more
constraint,
reports the corresponding
first
in
1975. Constraint on E.xecutive
from the Polity
stage. Panel
C
111
data
set.
reports the
Panel
OLS
in
A
1970
is
measured on
reports the
coefficient
two stage
America
is
in
column
parentheses. In regressions with continent dummies, the
omitted.
See Appendix Table Al for more detailed variable definitions and sources.
1
GDP pc
from regressing log
capita on average protection against expropriation risk, with the other control variables indicated in that
not reported to save space). Standard errors are
a scale of
least squares
(fiill
dummy
results
for
Appendix Table A4a
GDP per capita
IV Regressions of log
Base
Base
Sample
Sample
(1)
(2)
Base
Base
Base
Base
Base
Base
Sample
Sample
Sample
Sample
Sample
Sample
with
with
without
without
Continent
Continent
without neo without neo-
Europes
Europes
Africa
Africa
Dummies
Dummies
(3)
(4)
(5)
(6)
(7)
(8)
Panel A: Two Stage Least Squares
Constraint on Executive
in
1990
0.51
0.47
0.46
0.45
0.77
0.67
0.69
0.62
(0.08)
(0.09)
(0.10)
(0.10)
(0.32)
(0.31)
(0.23)
(0.24)
0.82
Latitude
(1.00)
Asia
0.43
0.78
0.77
(1.10)
(1.60)
(1.2)
Dummy
Africa
Dummy
"Other" Continent
Dummy
Panel B: First-Stage for Constraint on Executive
Log
Mortality
0.31
(0.57)
0.89
0.70
(0.80)
(0.78)
-0.12
-0.12
(0.6)
(0.6)
1990
-1.1
-1
-1.0
-1.0
-0.86
-0.9
-0.63
-0.61
(0.19)
(0.21)
(0.20)
(0.22)
(0.38)
(0.40)
(0.21)
(0.22)
Latitude
Asia
in
0.31
(0.60)
1.30
0.67
-0.20
0.6
(2.00)
(2.50)
(2.50)
(1.8)
Dummy
Africa
-1.7
-1.7
(0.7)
(0.7)
-2.3
-2.3
(0.53)
(0.53)
-0.2
-0.14
(0.90)
(0.91)
0.48
0.47
Dummy
"Other" Continent
Dummy
Adjusted R-Squared
0.34
0.33
Constraint on Executive
0.29
0.25
0.24
0.23
0.23
0.2
0.19
0.16
(0.04)
(0.04)
(0.04)
(0.04)
(0.07)
(0.07)
(0.05)
(0.05)
69
69
65
65
35
35
69
69
0.24
0.25
0.08
0.11
Panel C: Ordinary Least Squares
in
1990
Number of Observations
Constraint on Executive in 1990
is
measured on a
arbitrary actions that the executive can take,
the
two stage
constraint
least squares estimates
on the executive using log
coefficient
from regressing log
indicated in that
column
continent dummies, the
from
to 7,
1
with log
for
with a
set (see
Appendix Table
GDP per capita (PPP basis)
settler mortality;
Panel
B
in
1
for
more
detail).
Panel
A reports
1995 as the dependent variable, instrumenting for
reports the corresponding
GDP per capita on constraint on
(full results
dummy
scale,
from the Polity data
higher score indicating more constraints on the
first stage.
Panel
C reports the OLS
the executive in 1990, with the other control variables
not reported to save space). Standard errors are in parentheses. In regressions with
America
is
omitted.
Appendix Table A4b
IV Regressions of log
GDP per capita
Base
Base
Sample
Sample
Base
Base
Base
Base
Sample
Sample
with
with
without
without
Sample
Sample
Base
Base
neo-
neo-
without
without
Continent Continent
Sample
Sample
Europes
Europes
Africa
Africa
Dummies Dummies
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Panel A: Two Stage Least Squares
Law and Order Tradition
in
1
995
1.1
1.2
1.3
1.4
0.72
0.72
1
1.10
(0.21)
(0.33)
(0.36)
(0.45)
(0.15)
(0.18)
(0.28)
(0.5)
Latitude
Asia
-1.6
-1.3
-0.39
-1.7
(1.80)
(2.00)
(1.10)
(1.9)
Dummy
Africa
Dummy
"Other" Continent
Dummy
Panel B: First-Stage for Law and Order Tradition
Log Mortality
-0.70
-0.73
(0.30)
(0.35)
-0.95
-1
(0.7)
(0.8)
-0.4
-0.35
-1
-0.91
-0.45
-0.33
(0.10)
(0.11)
(0.11)
(0.10)
(0.17)
(0.18)
(0.13)
(0.13)
2.40
1.50
1.40
2.3
(1.00)
(1.20)
(1.10)
(1.05)
Dummy
Adjusted R-Squared
(0.50)
-0.42
Dummy
"Other" Continent
(0.40)
1995
Dummy
Africa
-1.20
-0.53
Latitude
Asia
in
-1.1
0.17
0.4
0.5
(0.4)
(0.4)
-0.03
-0.01
(0.32)
(0.30)
1.2
1
(0.52)
(0.50)
0.34
0.38
0.3
0.35
0.5
0.42
0.41
0.37
0.51
0.49
0.4
0.35
(0.10)
(0.10)
(0.11)
(0.11)
(0.10)
(0.11)
(0.09)
(0.10)
63
63
59
36
63
0.51
0.5
0.1^
Panel C: Ordinary Least Squares
Law and Order Tradition
Number of Observations
"Law ana Ureter tradition
Political
log
1995
in
in
IVVS"
is
measured on a scale trom U
Risk Services (see Appendix Table Al for more
GDP per capita (PPP basis)
settler mortality;
Panel
B
in
detail).
36
59
to 0,
Panel
where
A
a tiigher score
reports the
means more law and
two stage
least
1995 as the dependent variable and instrumenting for law and order tradition using log
reports the corresponding
first stage.
Panel
C
reports the
OLS
coefficient from regressing log
(full results
reported to save space). Standard errors are in parentheses. In regressions with continent dummies, the
omitted.
63
trom
squares estimates with
per capita on law and order tradition in 1995, with the other control variables indicated in that column
is
order,
dummy
GDP
not
for
America
Appendix Table A4c
IV Regressions of log
Base
Base
Sample
Base
Base
Sample
Sample
neo-
without
without
Continent Continent
Sample
Sample
Europes
Europes
Africa
Africa
Dummies Dummies
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
:
Two Stage Least Squares
-L6
-1.8
-2.1
-2.3
-0.9
-0.9
-1.3
-1.50
(0.33)
(0.56)
(0.80)
(1.00)
(0.15)
(0.18)
(0.46)
(0.7)
-2.1
-1.3
-0.26
-1.4
(2.20)
(2.70)
(0.90)
(1.9)
Panel B: First-Stage for Property Rights
0.37
0.28
0.2
(0.10)
(0.10)
(0.10)
Latitude
Asia
with
without
neo-
Dummy
Log Mortality
with
without
Dummy
"Other" Continent
Base
Sample
Base
Dummy
Africa
Base
Sample
Base
Latitude
Asia
per capita
Sample
Panel A
Property Rights in 1997
GDP
in
(0.17)
(0.11)
-0.09
(0.45)
-0.44
-0.47
(0.44)
(0.49)
-0.28
-0.21
(0.6)
(0.6)
1997
0.75
0.21
-0.14
(0.42)
0.67
0.31
0.24
(0.19)
(0.13)
(0.13)
-2.00
-1.10
-1.20
-1.8
(1.00)
(1.20)
(1.20)
(1.00)
Dummy
Africa
0.5
0.4
(0.4)
(0.4)
Dummy
"Other" Continent
Dummy
Adjusted R-Squared
0.06
0.3
0.25
(0.32)
(0.31)
-0.24
-0.14
(0.47)
(0.47)
0.17
0.2
0.17
0.21
-0.69
-0.6
-0.6
-0.6
-0.65
-0.63
-0.51
-0.5
(0.10)
(0.10)
(0.10)
(0.10)
(0.08)
(0.08)
(0.08)
(0.08)
69
69
0.06
0.3
0.3
Panel C: Ordinary' Least Squares
Property Rights in
1
997
Number of Observations
Troperty Rights
m
1
vys"
is
69
measured on
a scale
69
trom
to 5,
where
from the Heritage Foundation (see Appendix Table Al for more
estimates with log
GDP per capita (PPP basis)
log settler mortality; Panel
GDP per capita on
B
in
1
reports the corresponding
as the
first
41
a higher score
detail).
Panel
means
A reports
less protection tor
the
two stage
least
property nghts,
squares
dependent variable and instrumenting for property rights using
stage.
Panel
C
reports the
OLS
coefficient
property rights in 1995, with the other control variables indicated in that column
to save space). Standard errors are in parentheses.
omitted.
995
65
65
1
In regressions with continent
dummies,
the
from regressing log
(full results
dummy
for
not reported
America
is
Appendix Table A4d
GDP
IV Regressions of log
per capita
Base
Base
Sample
Sample
Base
Base
Base
Base
Sample
Sample
with
with
without
without
Sample
Sample
Base
Base
neo-
neo-
without
without
Continent Continent
Sample
Sample
Europes
Europes
Africa
Africa
Dummies Dummies
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Pane! A: Two Stage Least Squares
Rule of Law
in
1995
0.59
0.77
(0.14)
(0.30)
1.0
(0.50)
(0.31)
0.35
0.51
0.56
0.71
(0.12)
(0.27)
(0.19)
(0.35)
-3.6
-3.1
-3.9
-2.4
(3.30)
(4.20)
(3.50)
(3.0)
Latitude
Asia
0.8
Dummy
Africa
Dummy
"Other" Continent
Dummy
Panel B: First-Stage for Rule of Law
Log
-0.37
(0.50)
(0.63)
-1.5
-1.8
(0.9)
(1.3)
1995
-0,75
-0.61
-1.7
-1.1
-1
-0.77
(0.27)
(0.31)
(0.30)
(0.32)
(0.56)
(0.58)
(0.35)
(0.39)
5.70
3.70
8.80
4.2
(2.80)
(3.30)
(3.60)
(2.90)
Dummy
"Other" Continent
-0.41
-0.8
Dummy
Africa
0.32
(0.77)
-1.1
Mortality
Latitude
Asia
in
0.29
(0.62)
Dummy
Adjusted R-Squared
0.08
-1.7
-1.3
(1.0)
(1.1)
-0.31
-0.3
(0.83)
(0.80)
2.4
2.2
(1.30)
(1.30)
0.28
0.25
0.27
0.2
0.24
0.18
0.13
0.14
0.12
0.14
0.13
0.13
0.1
(0.04)
(0.04)
(0.04)
(0.04)
(0.04)
(0.05)
(0.04)
(0.04)
38
38
65
65
0.09
0.18
Panel C: Ordinary Least Squares
Rule of Law
1995
in
Number of Observations
"Rule of Law
Institute (see
capita
(PPP
in
1995"
is
Appendix Table Al
basis) in
for
65
a scale
more
from
detail).
61
61
to 10,
Panel
A
where
a higher score
reports the
two stage
means stronger
least
first
stage.
Panel
C
reports the
1995, with the other control variables indicated
are in parentheses. In regressions with continent
in that
OLS
coefficient
column
dummies, the
from regressing log
(full results
dummy
for
rule of law,
from the Fraser
squares estimates with log
1995 as the dependent variable and instrumenting for rule of law using log
reports the corresponding
m
65
measured on
settler mortality;
GDP per capita
GDP per
Panel
B
on rule of law
not reported to save space). Standard errors
America
is
omitted.
Appendix Table A4e
IV Regressions of log
GDP per capita
Base
Base
Sample
Sample
Base
Base
Base
Base
Sample
Sample
with
with
without
without
Sample
Sample
Base
Base
neo-
neo-
without
without
Continent Continent
Sample
Sample
Europes
Europes
Africa
Africa
Dummies Dummies
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Panel A: Two Stage Least Squares
0.61
0.61
0.73
0.7
0.39
0.36
0.63
0.65
(0.13)
(0.15)
(0.25)
(0.24)
(0.08)
(0.09)
(0.15)
(0.2)
Efficiency of the Judiciary
1980-83
Latitude
Asia
-0.08
0.64
0.64
-0.24
(1.20)
(1.60)
(0.90)
(1.3)
Dummy
Africa
Dummy
"Other" Continent
Dummy
-0.78
-0.81
(0.37)
(0.42)
-0.43
-0.43
(0.38)
(0.40)
-1.3
-1.4
(0.7)
(0.7)
Panel B: First-Stage for Efficiency of the Judiciary 1980-83
Log Mortality
-1
-1
-0.74
-0.79
-1.8
-1.9
-0.94
-0.94
(0.23)
(0.27)
(0.28)
(0.30)
(0.28)
(0.32)
(0.26)
(0.31)
Latitude
Asia
0.70
-1.30
-1.20
-1.06
(2.10)
(2.60)
(1.80)
(2.20)
Dummy
Africa
Dummy
"Other" Continent
Dummy
-0.01
0.001
(0.7)
(0.7)
-0.13
-0.13
(0.70)
(0.70)
2.2
2.2
(1.00)
(1.00)
0.37
0.35
Adjusted R-Squared
0.34
0.32
Efficiency of the Judiciary
0.34
0.31
0.28
0.3
0.33
0.32
0.32
0.3
(0.06)
(0.06)
(0.07)
(0.07)
(0.06)
(0.07)
(0.06)
(0.06)
Number of Observations
38
38
34
27
27
38
38
"Efficiency of the Judiciary 1980-83"
is
0.14
0.16
0.59
0.6
Panel C: Ordinary Least Squares
1980-83
judiciary,
from Business International (see Appendix Table
estimates with log
GDP per capita
(PPP
GDP
America
is
1
for
to 10,
more
where
detail).
a higher score
Panel
means
A reports
the
a
more
two stage
efficient
least squares
B
reports the
coaesponding
first
stage. Panel
C
reports the
OLS
coefficient
per capita on efficiency of the judiciary, with the other control variables indicated in that column
results not reported to save space).
for
A
1
basis) in 1995 as the dependent variable and instrumenting for efficiency of the
judiciary using log settler mortality; Panel
regressing log
34
measured on a scale from
omitted.
Standard errors are
in parentheses.
In regressions with continent
dummies, the
from
(full
dummy
Appen dix Table A5
Constructi on of Europ ean Settlements Variable
Settlers/total
population
Country
Population of
in
1900 (low
European
estimate)
descent
in
1975 Notes and Sources
(page references are
to
McEvedy and Jones, unless otherwise
noted)
North America
pp. 283-284, indigenous people
Canada
0.99
0.98 and 0.25m
in
1975, out of
pp. 286-290; African
USA
0.88
total
0.1m
Americans were
1925, while
in
1975, Amerindians were 0.6m
population
0.84 Mexico were 8.75m,
when
total
1900, out of
9m
was 76m
in
total
in
total
population of 5.25m,
population of 23m.
in
1900 and Amerindians were 0.5m
in
in
1900, African Americans were
25m
1975, and immigrants from Puerto Rico and
population
was 210m
Central America and IVIexico
p. 291;
there
2m Spaniards in a total population of 13.5m in 1900; the
55% Mestizo (mixed race), 30% Amerindian, and 15% white
were
proportions of
Mexico
0.15
0.15 have stayed "remarkably constant"
Central America
0.20
0.20 pp.294; 3/5 Amerindian; 1/5 Spanish; 1/5 mestizo
Guatemala
0.20
0.23 assuming 1900 proportions
0.20
0.20 pp.294-296; assuming unchanged from Central American proportions of 1800
Honduras
0.20
0.20 pp.294-296; assuming unchanged from Central American proportions of 1800
Nicaragua
0.20
0.20 pp.294-296; assuming unchanged from Central American proportions of 1800
Costa Rica
0.20
0.20 pp.294-296; assuming unchanged from Central American proportions of 1800
Panama
0.20
0.20 pp.294-296; assuming unchanged from Central American proportions of 1800
Belize
0.20
0.20 pp.294-296; assuming unchanged from Central American proportions of 1800
0.10
0.10 1975
Jamaica
0.02
0.10 African" population
Haiti
0.00
0.00
Dominican Republic
Lesser Antilles
0.25
0.25 1975
Barbados
0.20
0.20 1975
0.40
0.40 1975
0.20
0.25 1900 to 1975
0.20
0.20
pp.294-296;
El
Salvador
in
1975
55%
Mestizo,
same
in
1800
22.5% Amerindians, and 22.5% white;
for Central America as a whole in 1800
as
Caribbean Islands
p.300, using 1975 non-African proportion; assuming no change from 1900 to
Bahamas
Greater Antilles
Rogozinski (1992), p.186 for 1911; 1975 from McEvedy and Jones p.300, "non
Rogozinski (1992), p.217 on expulsion of settlers from 1805; 1975 from
McEvedy and Jones p.300, using "non-African population"
p.300, using 1975 non-African proportion; assuming no
change from 1900
to
p.300, using 1975 non-African proportion; assuming no change from 1900 to
p.300, using 1975 non-African proportion; assuming no change from 1900 to
Trinidad and
Tobago
South America
p.302; assuming
Colombia
Venezuela
p.
16%
0.01
0.01
Guyana
0.02
0.02 1975
0.27
0.30
Pern
0.30
0.30
Ecuador
0,30
0.30
Andean Region
(ex-lnca)
in
1975, o.4m
Indonesian,
2%
total
population, which
Chinese,
1%
was 40%,
Amerindian, and
black, 40%, Asian Indian,
1%
whites were 25% of the population in 1822 at time of independence; by
1975 had increase to 55%; assuming half this increased in proportion had
occun-edby 1900
by 1900 there were 2m of Spanish descent in the region, with a total population
of 7.35m; in 1975 there were 9m of Spanish descent, with a total population of
30.25
p.312; whites have consistently been 30% of the population; with 30% Mestizo
and 40% Amerindian
p. 31 2; whites have consistently been 30% of the population; with 30% Mestizo
and 40%, Amerindian
p. 308;
0.55
change from
white; assuming no
change from 1900 to 1975
p.304; in 1975, 0.8m population, 50% Asian Indian, 30% black, 4%
Amerindian. 1% Chinese and 2% white; assuming no change from 1900 to
Surinam
0.40
ethnic mix as Venezuela; assuming no
302; assuming no change from 1900 to 1975
p.304;
Brazil
same
Appen dix Table A5
Constructi on of Europ ean Settlements Variable
Settlers/total
population
Country
Population of
in
1900 (low
European
estimate)
descent
Bolivia
0.30
Paraguay
0.25
in
1975 Notes and Sources
p.312; whites have consistently been 30% of the population; with 30% Mestizo
0.30 and 40% Amerindian
p.31 2; only 30,000 Amerindians in 1 975, out of total population of 2.5m, 75%
mestizos and 25% white; two thirds of adult male population died or
disappeared due to war in 1865-1870; assuming no change in composition of
0.25 population from 1900 to 1975
30% of population was foreign born in 1914 census (p.313). in 1975
10% of population had Indian or mixed ancestry; white population was
p. 313-4;
only
1 5m, Indians were 0.35m, and Mestizos were 0-75m in 1 825 (for Argentina
and Chile combined); most immigrants arrived after 1880s, with peak years in
1910s. White population was 15m out of total of 17m in 1950. Population was
0.
Argentina
0.90 4.75m
0.60
in
1900;
we assume 50%
white population
in
1900 (above
that of Chile).
P-313-4; never more than 5% foreign born in any census (p.313); in Chile,
population equally divided in 1975 between whites and those of Indian or mixec
descent; since 1850 Argentina has received at least 2.5m net immigrants, while
Chile
0.50
0.50 Chile has only received 0.2m
Uruguay
0.60
0.90 p.313-4; assuming
Australia
0.98
0.99 aborigines out of population of 13.5m
New
0.93
0.92 250,000 Maori
same
ethnic mix as Argentina
Oceania
pp. 327-8; 60,000 aborigines out of population of
pp.337-8; 42,000 Maori
Zealand
in
"in
in
3.75m
in
1900; 80,000
1975.
the 1890s", total population of 750,000
1975, out of
total
in
1900;
population of 3m.
Africa
The Maghreb
Morocco
0.01
Algeria
0.13
Tunisia
0.00
•
0.00
assuming half of settlers in non-Algerian Mahgreb went to Morocco (p.220); In
1956 Europeans were 5.4 percent in former French Morocco (Curtin et al 1995,
p.435); most left by end of 1960s
p.220; In 1956 Europeans were 10.9 percent in Algeria (Curtin et al 1995,
p.435); most left by end of 1960s
assuming half of settlers in non-Algerian Mahgreb went to Tunisia; In 1956
Europeans were 6.7 percent (Curtin et al 1995, p.435), most left by end of
1960s
in 1956 no Europeans (Curtin el al 1995)
p. 226; in 1882 Egypt had a population of 6,800.000 of which 90,000 were
foreigners (Curtin 1998, p.127); most left by end of 1960s
p. 230; in 1956 no Europeans (Curtin et al 1995, p.435)
pp. 235-236; in 1956 no Europeans (Curtin et al 1995)
0.03
0.00
Libya
0.00
0.00
Egypt
0.01
0.00
Ethiopia
0.00
0.00
Sudan
0.00
0.00
Mauritania
0.00
0.00 pp.238-240,
in
Mali
0.00
0.00 pp.238-240;
in
Niger
0.00
0.00 pp.238-240;
in
Chad
0.00
0.00 pp.238-240,
in
Senegal
0.00
0.00 pp.241-246;
in
Gambia
0.00
in
Sahel States
West
1956
1956
1956
1956
no
no
no
no
Europeans
Europeans
Europeans
Europeans
(Curtin et
1956
1956
1956
1956
1956
1956
1956
1956
1956
1956
1956
1956
no
no
no
no
no
no
no
no
no
no
no
no
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
(Curtin et al 1995)
1956
1956
1956
1956
1956
1956
no
no
no
no
no
no
Europeans
Europeans
Europeans
Europeans
Europeans
Europeans
al
1995)
(Curtin et al 1995)
(Curtin et al 1995)
(Curtin et
al
1995)
Africa
Guinea-Bissau
0.00
0.00 pp. 241-246;
0.00 pp.241-246;
Guinea-Conaltry
0.00
0.00 pp.241-246;
in
Siena Leone
0.00
0.00 pp.241-246;
in
0.00
0.00 pp.241-246;
in
Liberia
Coast
in
0.00
0.00 pp.241-246;
in
Ghana
Togo
0.00
0.00 pp.241-246;
in
0.00
0.00 pp.241-246,
in
Benin
0.00
0.00 pp.241-246;
in
Nigeria
0.00
0.00 pp.241-246;
in
Upper Volta
0.00
0.00 pp.241-246;
in
Ivory
(Curtin et
al
1995)
(Curtin et
al
1995)
(Curtin et al 1995)
(Curtin et
al
1995)
(Curtin et al 1995)
(Curtin et al
1995)
(Curtin et al 1995)
(Curtin etal
1995)
(Curtin et al 1995)
(Curtin et al 1995)
(Curtin et al 1995)
Central-West Africa
0.00
0.00 pp.247-248;
in
Cameroon
0.00
in
Central African Republic
0.00
0.00 pp. 247-248;
0.00 pp.247-248;
Gabon
Congo
0.00
0.00 pp.247-248;
in
0.00
0.00 pp.247-248;
in
Equatorial Guinea
0.00
0.00 pp.247-248:
in
0.01
0.00 Using 1956 estimate, from Curtin et
Equatoria
Zaire
in
(Curtin et al 1995)
(Curtin et al 1995)
(Curtin et al 1995)
(Curtin et al 1995)
(Curtin et al 1995)
(Curtin et al 1995)
al
(1995)
Appen dix Table A5
Constructi on of Europ ean Settlements Variable
Settlers/total
population
Country
Population of
in
1900 (low
European
estimate)
descent
Angola
in
1975 Notes and Sources
500,000 settlers left before and
0.08
0.00 2.5 percent (Curtin et
al
after
independence;
in
1956 Europeans were
1995)
East Africa
1956 no Europeans (Curtin
tjganda
0.00
0.00 pp. 250-252;
Kenya
0.01
0.00 Using 1956 estimate, from Curtin et
Tanzania
0.00
0.00 pp.250-252;
in
Rwanda and Burundi
0.00
0.00 pp.250-252;
in
Zambia
0.03
0.00 1960s
Rhodesia
0.07
0.04 0.23m settlers, while
Malawi
0.03
0.00 1960s
0.03
0.00 p.256;
in
al
et al 1995)
(1995)
1956 no Europeans (Curtin
1956 no Europeans (Curtin
et al 1995)
et al 1995)
South-Central Africa
Using 1956 estimate, from Curtin et
Using 1956 estimate, from Curtin
total
al
most
settlers left
by end of
1900; in 1965 there were
1975 was 6.25m
(1995); most settlers left by end of
et al (1995), for
population
Using 1956 estimate, from Curtin
(1995);
in
et al
Soutliem Africa
Mozambique
0.22
all 150,000 Portuguese settlers left after independence
1956 Europeans were 20.8 percent and "coloreds" were another 10 percent
(Curtin et al 1995); assuming no change in proportions from 1900 to 1975;
major immigration from Europe after 1867 and 1886, brought white population
to about 1/5 of total in South Africa; in 1900 there were 1.2m whites, 3.75m
Bantu, 0.4m "Cape coloureds", and 0.1m Indians; in 1975 there were 4.1m
whites, 2.3m "Cape coloureds", 0.75m Indians, and 18m blacks (plus 1.5m
0.16 Bantu in Swaziland and Lesotho)
Madagascar
0.00
0.00 pp.264-266; no Europeans
Mauritius
0.05
0.17 p.267;
Reunion
0.05
in 1956 (Curtin et
1800 5/6 of population were slaves
0.33 p.266; 3/4 of population slaves in 1300
Bangladesh
0.00
0.00 pp.182-186; no significant European settlements
India
0.00
0.00 pp.182-186; no significant European settlements
Pakistan
0.00
0.00 pp.182-186: no significant European settlements
0.00
0.00
0.00
0.00 p.190; no significant European settlements
0.00
0.00
Malaysia
0.00
0.00 these are not mentioned by
Singapore
0.05
Indonesia
0.00
In
South
Africa.
Swaziland and
Lesotho
Islands of the Western Indian
Ocean
al
1995, p.435)
in
Pakistan, India and Bangladesh
Sri
Lanka
Burma
p. 186;
no
significant
European settlements
Indo-China
p.196; there
Vietnam
were a few thousand settlers, but these are not mentioned by
all had left by 1975
McEvedy and Jones; almost
Malay Archipelago
there must have
been a few thousand settlers (rubber plantations), but
McEvedy and Jones; almost all had left by 1975
0.00 Assuming population composition similar to Hong Kong
p. 198; there were a few settlers, but these are not mentioned by McEvedy and
0.00 Jones
p. 198;
from 1895 census; proportion of non-Chinese peaked
Hong Kong
0.04
Welsh (1993), p.253; population was
0.00 subjects who were European in 1931
to McEvedy and Jones, unless otherwise noted.
Other sources; Rogozinski (1994), Welsh (1994), McEvedy and Jones (1975), Curtin
4m
in
at
0.06
in
1869; from
1971; there were 6,636 British
Page references are
et al (1995,
p.435 unless otherwise indicated).
un
CD
I
CX3
o
'i—
Q.
O
Q.
X
LU
CD
C
"03
<
c
o
o
CD
CD
CD
>
<
C\J
T
o
CO
9661
CO
'dcdcd 'Biideo
jed
dOO
Boi
Average Protection Against Expropriation Risk 1985-95
CD
00
-J.
M
>
l_
in
T'
M
r-
(
en
a
-0
>
<r
rn
> >
^
00
D
O
CD
CQ
X
a-,
N
>
73
CD.
00
to
I
CNJ
c
o
"to
v>
o
O)
0)
E
o
00 i:
ro
-1^
'5.
(0
u
•
a
Q
o
Q.
1—
**-
o
0)
>
o
a>
T3
1^
00
CM
9661-
LJ!
CD
(ddd) BjidBO jad
dOO
Bo| lenjov
5'^
ne [ks
Date Due
Lib-26-67
MIT LIBRARIES
3 9080 01917 6038