reversal of fortune: geography and daron...
Post on 01-Aug-2020
3 Views
Preview:
TRANSCRIPT
REVERSAL OF FORTUNE: GEOGRAPHY ANDINSTITUTIONS IN THE MAKING OF THE MODERN
WORLD INCOME DISTRIBUTION*
DARON ACEMOGLU
SIMON JOHNSON
JAMES A. ROBINSON
Among countries colonized by European powers during the past 500 years,those that were relatively rich in 1500 are now relatively poor. We document thisreversal using data on urbanization patterns and population density, which, weargue, proxy for economic prosperity. This reversal weighs against a view thatlinks economic development to geographic factors. Instead, we argue that thereversal reflects changes in the institutions resulting from European colonialism.The European intervention appears to have created an “institutional reversal”among these societies, meaning that Europeans were more likely to introduceinstitutions encouraging investment in regions that were previously poor. Thisinstitutional reversal accounts for the reversal in relative incomes. We providefurther support for this view by documenting that the reversal in relative incomestook place during the late eighteenth and early nineteenth centuries, and resultedfrom societies with good institutions taking advantage of the opportunity toindustrialize.
I. INTRODUCTION
This paper documents a reversal in relative incomes amongthe former European colonies. For example, the Mughals in Indiaand the Aztecs and Incas in the Americas were among the richestcivilizations in 1500, while the civilizations in North America,New Zealand, and Australia were less developed. Today theUnited States, Canada, New Zealand, and Australia are an orderof magnitude richer than the countries now occupying the terri-tories of the Mughal, Aztec, and Inca Empires.
* We thank Joshua Angrist, Abhijit Banerjee, Olivier Blanchard, AlessandraCassella, Jan de Vries, Ronald Findlay, Jeffry Frieden, Edward Glaeser, HerschelGrossman, Lawrence Katz, Peter Lange, Jeffrey Sachs, Andrei Shleifer, FabrizioZilibotti, three anonymous referees, and seminar participants at the All-Univer-sities of California History Conference at Berkeley, the conference on “Globaliza-tion and Marginalization” in Bergen, The Canadian Institute of Advanced Re-search, Brown University, the University of Chicago, Columbia University, theUniversity of Houston, Indiana University, Massachusetts Institute of Technol-ogy, National Bureau of Economic Research summer institute, Stanford Univer-sity, the Wharton School of the University of Pennsylvania, and Yale Universityfor useful comments. Acemoglu gratefully acknowledges financial help from TheCanadian Institute for Advanced Research and the National Science FoundationGrant SES-0095253. Johnson thanks the Massachusetts Institute of TechnologyEntrepreneurship Center for support.
© 2002 by the President and Fellows of Harvard College and the Massachusetts Institute ofTechnology.The Quarterly Journal of Economics, November 2002
1231
Our main measure of economic prosperity in 1500 is urban-ization. Bairoch [1988, Ch. 1] and de Vries [1976, p. 164] arguethat only areas with high agricultural productivity and a devel-oped transportation network can support large urban popula-tions. In addition, we present evidence that both in the timeseries and the cross section there is a close association betweenurbanization and income per capita.1 As an additional proxy forprosperity we use population density, for which there are rela-tively more extensive data. Although the theoretical relationshipbetween population density and prosperity is more complex, itseems clear that during preindustrial periods only relativelyprosperous areas could support dense populations.
With either measure, there is a negative association betweeneconomic prosperity in 1500 and today. Figure I shows a negativerelationship between the percent of the population living in townswith more than 5000 inhabitants in 1500 and income per capitatoday. Figure II shows the same negative relationship betweenlog population density (number of inhabitants per square kilome-ter) in 1500 and income per capita today. The relationships shownin Figures I and II are robust—they are unchanged when wecontrol for continent dummies, the identity of the colonial power,religion, distance from the equator, temperature, humidity, re-sources, and whether the country is landlocked, and when weexclude the “neo-Europes” (the United States, Canada, New Zea-land, and Australia) from the sample.
This pattern is interesting, in part, because it provides anopportunity to distinguish between a number of competing theo-ries of the determinants of long-run development. One of the mostpopular theories, which we refer to as the “geography hypothe-sis,” explains most of the differences in economic prosperity bygeographic, climatic, or ecological differences across countries.The list of scholars who have emphasized the importance ofgeographic factors includes, inter alia, Machiavelli [1519], Mon-
1. By economic prosperity or income per capita in 1500, we do not refer to theeconomic or social conditions or the welfare of the masses, but to a measure oftotal production in the economy relative to the number of inhabitants. Althoughurbanization is likely to have been associated with relatively high output percapita, the majority of urban dwellers lived in poverty and died young because ofpoor sanitary conditions (see, for example, Bairoch [1988, Ch. 12]).
It is also important to note that the Reversal of Fortune refers to changes inrelative incomes across different areas, and does not imply that the initial in-habitants of, for example, New Zealand or North America themselves becamerelatively rich. In fact, much of the native population of these areas did notsurvive European colonialism.
1232 QUARTERLY JOURNAL OF ECONOMICS
tesquieu [1748], Toynbee [1934–1961], Marshall [1890], andMyrdal [1968], and more recently, Diamond [1997] and Sachs[2000, 2001]. The simplest version of the geography hypothesisemphasizes the time-invariant effects of geographic variables,such as climate and disease, on work effort and productivity, andtherefore predicts that nations and areas that were relatively richin 1500 should also be relatively prosperous today. The reversalin relative incomes weighs against this simple version of thegeography hypothesis.
More sophisticated versions of this hypothesis focus on thetime-varying effects of geography. Certain geographic character-istics that were not useful, or even harmful, for successful eco-nomic performance in 1500 may turn out to be beneficial later on.A possible example, which we call “the temperate drift hypothe-sis,” argues that areas in the tropics had an early advantage, butlater agricultural technologies, such as the heavy plow, croprotation systems, domesticated animals, and high-yield crops,have favored countries in the temperate areas (see Bloch [1966],Lewis [1978], and White [1962]; also see Sachs [2001]). Althoughplausible, the temperate drift hypothesis cannot account for the
FIGURE ILog GDP per Capita (PPP) in 1995 against Urbanization Rate in 1500
Note. GDP per capita is from the World Bank [1999]; urbanization in 1500 ispeople living in towns with more than 5000 inhabitants divided by total popu-lation, from Bairoch [1988] and Eggimann [1999]. Details are in Appendices 1and 2.
1233REVERSAL OF FORTUNE
reversal. First, the reversal in relative incomes seems to be re-lated to population density and prosperity before Europeans ar-rived, not to any inherent geographic characteristics of the area.Furthermore, according to the temperate drift hypothesis, thereversal should have occurred when European agricultural tech-nology spread to the colonies. Yet, while the introduction of Eu-ropean agricultural techniques, at least in North America, tookplace earlier, the reversal occurred during the late eighteenth andearly nineteenth centuries, and is closely related to industrializa-tion. Another version of the sophisticated geography hypothesiscould be that certain geographic characteristics, such as the pres-ence of coal reserves or easy access to the sea, facilitated indus-trialization (e.g., Pomeranz [2000] and Wrigley [1988]). But we donot find any evidence that these geographic factors caused indus-trialization. Our reading of the evidence therefore provides littlesupport to various sophisticated geography hypotheses either.
An alternative view, which we believe provides the best ex-planation for the patterns we document, is the “institutions hy-pothesis,” relating differences in economic performance to theorganization of society. Societies that provide incentives and op-portunities for investment will be richer than those that fail to doso (e.g., North and Thomas [1973], North and Weingast [1989],
FIGURE IILog GDP per Capita (PPP) against Log Population Density in 1500
Note. GDP per capita from the World Bank [1999]; log population density in1500 from McEvedy and Jones [1978]. Details are in Appendix 2.
1234 QUARTERLY JOURNAL OF ECONOMICS
and Olson [2000]). As we discuss in more detail below, we hy-pothesize that a cluster of institutions ensuring secure propertyrights for a broad cross section of society, which we refer to asinstitutions of private property, are essential for investment in-centives and successful economic performance. In contrast, ex-tractive institutions, which concentrate power in the hands of asmall elite and create a high risk of expropriation for the majorityof the population, are likely to discourage investment and eco-nomic development. Extractive institutions, despite their adverseeffects on aggregate performance, may emerge as equilibriuminstitutions because they increase the rents captured by thegroups that hold political power.
How does the institutions hypothesis explain the reversal inrelative incomes among the former colonies? The basic idea isthat the expansion of European overseas empires starting at theend of the fifteenth century caused major changes in the organi-zation of many of these societies. In fact, historical and econo-metric evidence suggests that European colonialism caused an“institutional reversal”: European colonialism led to the develop-ment of institutions of private property in previously poor areas,while introducing extractive institutions or maintaining existingextractive institutions in previously prosperous places.2 Themain reason for the institutional reversal is that relatively poorregions were sparsely populated, and this enabled or inducedEuropeans to settle in large numbers and develop institutionsencouraging investment. In contrast, a large population and rela-tive prosperity made extractive institutions more profitable forthe colonizers; for example, the native population could be forcedto work in mines and plantations, or taxed by taking over existingtax and tribute systems. The expansion of European overseasempires, combined with the institutional reversal, is consistentwith the reversal in relative incomes since 1500.
Is the reversal related to institutions? We document that thereversal in relative incomes from 1500 to today can be explained,
2. By the term “institutional reversal,” we do not imply that it was societieswith good institutions that ended up with extractive institutions after Europeancolonialism. First, there is no presumption that relatively prosperous societies in1500 had anything resembling institutions of private property. In fact, theirrelative prosperity most likely reflected other factors, and even perhaps geo-graphic factors. Second, the institutional reversal may have resulted more fromthe emergence of institutions of private property in previously poor areas thanfrom a deterioration in the institutions of previously rich areas.
1235REVERSAL OF FORTUNE
at least statistically, by differences in institutions across coun-tries. The institutions hypothesis also suggests that institutionaldifferences should matter more when new technologies that re-quire investments from a broad cross section of the society be-come available. We therefore expect societies with good institu-tions to take advantage of the opportunity to industrialize, whilesocieties with extractive institutions fail to do so. The data sup-port this prediction.
We are unaware of any other work that has noticed or docu-mented this change in the distribution of economic prosperity.Nevertheless, many historians emphasize that in 1500 the Mu-ghal, Ottoman, and Chinese Empires were highly prosperous, butgrew slowly during the next 500 years (see the discussion andreferences in Section III).
Our overall interpretation of comparative development in theformer colonies is closely related to Coatsworth [1993] and En-german and Sokoloff [1997, 2000], who emphasize the adverseeffects of the plantation complex in the Caribbean and CentralAmerica working through political and economic inequality,3 andto our previous paper, Acemoglu, Johnson, and Robinson [2001a].In that paper we proposed the disease environment at the timeEuropeans arrived as an instrument for European settlementsand the subsequent institutional development of the former col-onies, and used this to estimate the causal effect of institutionaldifferences on economic performance. Our thesis in the currentpaper is related, but emphasizes the influence of population den-sity and prosperity on the policies pursued by the Europeans (seealso Engerman and Sokoloff [1997]). In addition, here we docu-ment the reversal in relative incomes among the former colonies,show that it was related to industrialization, and provide evi-dence that the interaction between institutions and the opportu-nity to industrialize during the nineteenth century played a cen-tral role in the long-run development of the former colonies.4
3. In this context, see also Frank [1978], Rodney [1972], Wallerstein [1974–1980], and Williams [1944].
4. Our results are also relevant to the literature on the relationship betweenpopulation and growth. The recent consensus is that population density encour-ages the discovery and exchange of ideas, and contributes to growth (e.g., Boserup[1965], Jones [1997], Kremer [1993], Kuznets [1968], Romer [1986], and Simon[1977]). Our evidence points to a major historical episode of 500 years where highpopulation density was detrimental to economic development, and therefore shedsdoubt on the general applicability of this recent consensus.
1236 QUARTERLY JOURNAL OF ECONOMICS
The rest of the paper is organized as follows. The next sectiondiscusses the construction of urbanization and population densitydata, and provides evidence that these are good proxies for eco-nomic prosperity. Section III documents the “Reversal of For-tune”—the negative relationship between economic prosperity in1500 and income per capita today among the former colonies.Section IV discusses why the simple and sophisticated geographyhypotheses cannot explain this pattern, and how the institutionshypothesis explains the reversal. Section V documents that thereversal in relative incomes reflects the institutional reversalcaused by European colonialism, and that institutions startedplaying a more important role during the age of industry. SectionVI concludes.
II. URBANIZATION AND POPULATION DENSITY
II.A. Data on Urbanization
Bairoch [1988] provides the best single collection and assess-ment of urbanization estimates. Our base data for 1500 consist ofBairoch’s [1988] urbanization estimates augmented by the workof Eggimann [1999]. Merging the Eggimann and Bairoch seriesrequires us to convert Eggimann’s estimates, which are based ona minimum population threshold of 20,000, into Bairoch-equiva-lent urbanization estimates, which use a minimum populationthreshold of 5000. We use a number of different methods toconvert between the two sets of estimates, all with similar re-sults. Appendix 1 provides details about data sources and con-struction. Briefly, for our base estimates, we run a regression ofBairoch estimates on Eggimann estimates for all countries wherethey overlap in 1900 (the year for which we have most Bairochestimates for non-European countries). This regression yields aconstant of 6.6 and a coefficient of 0.67, which we use to generateBairoch-equivalent urbanization estimates from Eggimann’sestimates.
Alternatively, we converted the Eggimann’s numbers using auniform conversion rate of 2 as suggested by Davis’ and Zipf ’sLaws (see Appendix 1 and Bairoch [1988, Ch. 9]), and also testedthe robustness of the estimates using conversion ratios at theregional level based on Bairoch’s analysis. Finally, we con-structed three alternative series without combining estimatesfrom different sources. One of these is based on Bairoch, the
1237REVERSAL OF FORTUNE
second on Eggimann, and the third on Chandler [1987]. All fouralternative series are reported in Appendix 3, and results usingthese measures are reported in Table IV.
While the data on sub-Saharan Africa are worse than for anyother region, it is clear that urbanization in sub-Saharan Africabefore 1500 was at a higher level than in North America orAustralia. Bairoch, for example, argues that by 1500 urbaniza-tion was “well-established” in sub-Saharan Africa.5 Becausethere are no detailed urbanization data for sub-Saharan Africa,we leave this region out of the regression analysis when we useurbanization data, although African countries are included in ourregressions using population density.
Table I gives descriptive statistics for the key variables ofinterest, separately for the whole world, for the sample of ex-colonies for which we have urbanization data in 1500, and for thesample of ex-colonies for which we have population density datain 1500. Appendix 2 gives detailed definitions and sources for thevariables used in this study.
II.B. Urbanization and Income
There are good reasons to presume that urbanization andincome are positively related. Kuznets [1968, p. 1] opens his bookon economic growth by stating: “we identify the economic growthof nations as a sustained increase in per-capita or per-workerproduct, most often accompanied by an increase in populationand usually by sweeping structural changes. . . . in the distribu-tion of population between the countryside and the cities, theprocess of urbanization.”
Bairoch [1988] points out that during preindustrial periods alarge fraction of the agricultural surplus was likely to be spent ontransportation, so both a relatively high agricultural surplus anda developed transport system were necessary for large urbanpopulations (see Bairoch [1988, Ch. 1]). He argues “the existenceof true urban centers presupposes not only a surplus of agricul-
5. Sahelian trading cities such as Timbuktu, Gao, and Djenne (all in modernMali) were very large in the middle ages with populations as high as 80,000. Kano(in modern Nigeria) had a population of 30,000 in the early nineteenth century,and Yorubaland (also in Nigeria) was highly urbanized with a dozen towns withpopulations of over 20,000 while its capital Ibadan possibly had 70,000 inhabit-ants. For these numbers and more detail, see Hopkins [1973, Ch. 2].
1238 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EI
DE
SC
RIP
TIV
ES
TA
TIS
TIC
S
Wh
ole
wor
ld(1
)
Bas
esa
mpl
efo
ru
rban
izat
ion
(2)
Bas
esa
mpl
efo
rpo
pula
tion
den
sity
(3)
Bel
owm
edia
nu
rban
izat
ion
in15
00(4
)
Abo
vem
edia
nu
rban
izat
ion
in15
00(5
)
Bel
owm
edia
npo
pula
tion
den
sity
in15
00(6
)
Abo
vem
edia
npo
pula
tion
den
sity
in15
00(7
)
Log
GD
Ppe
rca
pita
(PP
P)
in19
958.
38.
57.
98.
88.
18.
37.
5(1
.1)
(0.9
)(1
.0)
Urb
aniz
atio
nin
1995
53.0
57.5
45.4
64.9
49.7
53.5
36.7
(23.
8)(2
2.4)
(22.
2)U
rban
izat
ion
in15
007.
36.
46.
42.
410
.52.
39.
5(5
.0)
(5.0
)(5
.0)
Log
popu
lati
onde
nsi
tyin
1500
1.0
0.2
0.5
�0.
91.
4�
0.6
1.6
(1.6
)(1
.9)
(1.5
)P
opu
lati
onde
nsi
tyin
1500
9.2
6.3
4.8
1.2
11.7
0.8
9.1
(24.
3)(1
6.4)
(11.
7)L
ogpo
pula
tion
den
sity
in10
000.
60.
110.
08�
1.20
1.22
�0.
941.
04(1
.5)
(2.0
)(1
.5)
Ave
rage
prot
ecti
onag
ain
stex
prop
riat
ion
,19
85–1
995
7.1
6.9
6.5
7.5
6.3
6.8
6.2
(1.8
)(1
.5)
(1.4
)C
onst
rain
ton
the
exec
uti
vein
1990
3.6
4.9
3.7
5.1
4.6
4.0
3.5
(2.3
)(2
.1)
(2.3
)C
onst
rain
ton
the
exec
uti
vein
firs
tye
arof
inde
pen
den
ce
3.6
3.3
3.4
3.8
2.8
3.6
3.3
(2.4
)(2
.5)
(2.3
)
Eu
rope
anse
ttle
men
tsin
1900
29.6
23.2
12.5
30.5
6.0
18.7
4.7
(41.
7)(2
8.7)
(22.
1)N
um
ber
ofob
serv
atio
ns
162
4191
2120
4744
Sta
nda
rdde
viat
ion
sar
ein
pare
nth
eses
.Nu
mbe
rof
obse
rvat
ion
sva
ries
acro
ssro
ws
due
tom
issi
ng
data
.Th
efi
rst
thre
eco
lum
ns
repo
rtm
ean
valu
esfo
rth
esa
mpl
ein
dica
ted
atth
eh
ead
ofth
eco
lum
n.T
he
last
fou
rco
lum
ns
repo
rtm
ean
valu
esfo
rfo
rmer
colo
nie
sbe
low
and
abov
eth
em
edia
n,s
epar
atel
yfo
rth
eba
seu
rban
izat
ion
and
popu
lati
onde
nsi
tysa
mpl
es.
For
deta
iled
sou
rces
and
desc
ript
ion
sse
eA
ppen
dix
2.
1239REVERSAL OF FORTUNE
tural produce, but also the possibility of using this surplus intrade” [p. 11].6 See de Vries [1976, p. 164] for a similar argument.
We supplement this argument by empirically investigatingthe link between urbanization and income in Table II. Columns(1)–(6) present cross-sectional regressions. Column (1) is for 1900,the earliest date for which we have data on urbanization andincome per capita for a large number of countries. The regressioncoefficient, 0.038, is highly significant, with a standard error of0.006. It implies that a country with 10 percentage points higherurbanization has, on average, 46 percent (38 log points) greaterincome per capita (throughout the paper, all urbanization ratesare expressed in percentage points, e.g., 10 rather than 0.1—seeTable I). Column (2) reports a similar result using data for 1950.Column (3) uses current data and shows that even today there isa strong relationship between income per capita and urbanizationfor a large sample of countries. The coefficient is similar, 0.036,and precisely estimated, with a standard error of 0.002. Thisrelationship is shown diagrammatically in Figure III.
Below, we draw a distinction between countries colonized byEuropeans and those never colonized (i.e., Europe and non-Euro-pean countries not colonized by Western Europe). Columns (4) and(5) report the same regression separately for these two samples. Theestimates are very similar: 0.037 for the former colonies sample, and0.033 for the rest of the countries. Finally, in column (6) we addcontinent dummies to the same regression. This leads to only aslightly smaller coefficient of 0.030, with a standard error of 0.002.
Finally, we use estimates from Bairoch [1978, 1988] to con-struct a small unbalanced panel data set of urbanization andincome per capita from 1750 to 1913. Column (7) reports a re-
6. The view that urbanization and income (productivity) are closely related isshared by many other scholars. See Ades and Glaeser [1999], De Long andShleifer [1993], Tilly and Blockmans [1994], and Tilly [1990]. De Long andShleifer, for example, write “The larger preindustrial cities were nodes of infor-mation, industry, and exchange in areas where the growth of agricultural pro-ductivity and economic specialization had advanced far enough to support them.They could not exist without a productive countryside and a flourishing tradenetwork. The population of Europe’s preindustrial cities is a rough indicator ofeconomic prosperity” [p. 675].
A large history literature also documents how urbanization accelerated inEurope during periods of economic expansion (e.g., Duby [1974], Pirenne [1956],and Postan and Rich [1966]). For example, the period between the beginning ofthe eleventh and mid-fourteenth centuries is an era of rapid increase in agricul-tural productivity and industrial output. The same period also witnessed a pro-liferation of cities. Bairoch [1988], for example, estimates that the number of citieswith more than 20,000 inhabitants increased from around 43 in 1000 to 107 in1500 [Table 10.2, p. 159].
1240 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EII
UR
BA
NIZ
AT
ION
AN
DP
ER
CA
PIT
AIN
CO
ME
Cro
ss-s
ecti
onal
regr
essi
onin
1913
,al
lco
un
trie
s(1
)
Cro
ss-s
ecti
onal
regr
essi
onin
1950
,al
lco
un
trie
s(2
)
Cro
ss-s
ecti
onal
regr
essi
onin
1995
,al
lco
un
trie
s(3
)
Cro
ss-s
ecti
onal
regr
essi
onin
1995
,on
lyfo
rex
-col
onie
s(4
)
Cro
ss-s
ecti
onal
regr
essi
onin
1995
,n
ever
colo
niz
edco
un
trie
son
ly(5
)
Cro
ss-s
ecti
onal
regr
essi
onin
1995
,al
lco
un
trie
s,w
ith
con
tin
ent
dum
mie
s(6
)
Pan
elda
tase
tth
rou
gh19
13(7
)
Dep
end
ent
vari
able
islo
gG
DP
per
capi
ta
Urb
aniz
atio
n0.
038
0.02
60.
036
0.03
70.
033
0.03
00.
026
(0.0
06)
(0.0
02)
(0.0
02)
(0.0
03)
(0.0
07)
(0.0
02)
(0.0
04)
R2
0.69
0.57
0.63
0.69
0.34
0.68
0.93
Nu
mbe
rof
obse
rvat
ion
s22
128
162
9351
162
55
Sta
nda
rder
rors
are
inpa
ren
thes
es.L
ogG
DP
per
capi
tath
rou
gh19
13is
from
Bai
roch
[197
8].U
rban
izat
ion
ispe
rcen
tof
popu
lati
onli
vin
gin
tow
ns
wit
hat
leas
t50
00pe
ople
,fr
omB
airo
ch[1
988]
thro
ugh
1900
wit
hsu
pple
men
tary
sou
rces
asde
scri
bed
inA
ppen
dix
1.L
ogG
DP
per
capi
tain
1950
isfr
omM
addi
son
[199
5];t
his
regr
essi
onu
ses
urb
aniz
atio
nin
1960
from
the
Wor
ldB
ank’
sW
orld
Dev
elop
men
tIn
dic
ator
s[1
999]
.L
ogG
DP
per
capi
ta(P
PP
)an
dU
rban
izat
ion
data
for
1995
are
from
the
Wor
ldB
ank’
sW
orld
Dev
elop
men
tIn
dic
ator
s[1
999]
.P
opu
lati
onde
nsi
tyis
tota
lpo
pula
tion
divi
ded
byar
able
lan
dar
ea,
both
from
McE
vedy
and
Jon
es[1
978]
.F
orde
tail
edso
urc
esan
dde
scri
ptio
ns
see
App
endi
x2.
Th
eco
un
trie
san
dap
prox
imat
eye
ars
for
wh
ich
we
hav
eda
ta(u
sed
inth
eu
nba
lan
ced
pan
elre
gres
sion
inco
lum
n(7
))ar
eA
ust
rali
a(1
830,
1860
,an
d19
13),
Au
stri
a(1
830,
1860
,19
13),
Bel
giu
m(1
830,
1860
,19
13),
Bri
tain
(175
0,18
30,
1860
,19
13),
Bu
lgar
ia(1
860,
1913
),C
anad
a(1
830,
1860
,19
13),
Ch
ina
(183
0,18
60),
Den
mar
k(1
830,
1860
,19
13),
Fin
lan
d(1
830,
1860
,191
3),F
ran
ce(1
750,
1830
,186
0,19
13),
Ger
man
y(1
830,
1860
,191
3),G
reec
e(1
860,
1913
),In
dia
(183
0,19
13),
Ital
y(1
830,
1860
,191
3),J
amai
ca(1
830,
1913
),Ja
pan
(175
0,18
30,1
913)
,Net
her
lan
ds(1
830,
1860
,191
3),N
orw
ay(1
830,
1860
,191
3),P
ortu
gal(
1830
,186
0,19
13),
Rom
ania
(183
0,18
60,1
913)
,Ru
ssia
(175
0,18
30,1
860,
1913
),S
pain
(183
0,18
60,
1913
),S
wed
en(1
830,
1860
,19
13),
Sw
itze
rlan
d(1
830,
1860
,19
13),
Un
ited
Sta
tes
(175
0,18
30,
1860
,19
13),
and
Yu
gosl
avia
(183
0,18
60,
1913
).
1241REVERSAL OF FORTUNE
gression of income per capita on urbanization using this paneldata set and controlling for country and period dummies. Theestimate is again similar: 0.026 (s.e. � 0.004). Overall, we con-clude that urbanization is a good proxy for income.
II.C. Population Density and Income
The most comprehensive data on population since 1 A.D.come from McEvedy and Jones [1978]. They provide estimatesbased on censuses and published secondary sources. Whilesome individual country numbers have since been revised andothers remain contentious (particularly for pre-Columbian Meso-America), their estimates are consistent with more recent re-search (see, for example, the recent assessment by the Bureauof the Census, www.census.gov/ipc/www/worldhis.html). We useMcEvedy and Jones [1978] for our baseline estimates, and testthe effect of using alternative assumptions (e.g., lower or higherpopulation estimates for Mexico and its neighbors before thearrival of Cortes).
FIGURE IIILog GDP per Capita (PPP) in 1995 against the Urbanization Rate in 1995
Note. GDP per capita and urbanization are from the World Bank [1999]. Ur-banization is percent of population living in urban areas. The definition of urbanareas differs between countries, but the usual minimum size is 2000–5000 inhabi-tants. For details of definitions and sources for urban population in 1995, see theUnited Nations [1998].
1242 QUARTERLY JOURNAL OF ECONOMICS
We calculate population density by dividing total populationby arable land (also estimated by McEvedy and Jones). Thisexcludes primarily desert, inland water, and tundra. As much aspossible, we use the land area of a country at the date we areconsidering.
The theoretical relationship between population density andincome is more nuanced than that between urbanization andincome. With a similar reasoning, it seems natural to think thatonly relatively rich areas could afford dense populations (seeBairoch [1988, Ch. 1]). This is also in line with Malthus’ classicwork. Malthus [1798] argued that high productivity increasespopulation by raising birthrates and lowering death rates. How-ever, the main thrust of Malthus’ work was how a higher thanequilibrium level of population increases death rates and reducesbirthrates to correct itself.7 A high population could therefore bereflecting an “excess” of population, causing low income per cap-ita. So caution is required in interpreting population density as aproxy for income per capita.
The empirical evidence regarding the relationship betweenpopulation density and income is also less clear-cut than therelationship between urbanization and income. In Acemoglu,Johnson, and Robinson [2001b] we documented that populationdensity and income per capita increased concurrently in manyinstances. Nevertheless, there is no similar cross-sectional rela-tionship in recent data, most likely because of the demographictransition—it is no longer true that high population density isassociated with high income per capita because the relationshipbetween income and the number of children has changed (e.g.,Notestein [1945] or Livi-Bacci [2001]).
Despite these reservations, we present results using popula-tion density, as well as urbanization, as a proxy for income percapita. This is motivated by three considerations. First, popula-tion density data are more extensive, so the use of populationdensity data is a useful check on our results using urbanizationdata. Second, as argued by Bairoch, population density is closely
7. A common interpretation of Malthus’ argument is that these populationdynamics will force all countries down to the subsistence level of income. In thatcase, population density would be a measure of total income, but not necessarilyof income per capita, and in fact, there would be no systematic (long-run) differ-ences in income per capita across countries. We view this interpretation asextreme, and existing historical evidence suggests that there were systematicdifferences in income per capita between different regions even before the modernperiod (see the references below).
1243REVERSAL OF FORTUNE
related to urbanization, and in fact, our measures are highlycorrelated. Third, variation in population density will play animportant role not only in documenting the reversal, but also inexplaining it.
III. THE REVERSAL OF FORTUNE
III.A. Results with Urbanization
This section presents our main results. Figure I in the intro-duction depicts the relationship between urbanization 1500 andincome per capita today. Table III reports regressions document-ing the same relationship. Column (1) is our most parsimoniousspecification, regressing log income per capita in 1995 (PPP basis)on urbanization rates in 1500 for our sample of former colonies.The coefficient is �0.078 with a standard error of 0.026.8 Thiscoefficient implies that a 10 percentage point lower urbanizationin 1500 is associated with approximately twice as high GDP percapita today (78 log points � 108 percent). It is important to notethat this is not simply mean reversion—i.e., richer than averagecountries reverting back to the mean. It is a reversal. To illustratethis, let us compare Uruguay and Guatemala. The native popu-lation in Uruguay had no urbanization, while, according to ourbaseline estimates Guatemala had an urbanization rate of 9.2percent. The estimate in column (1) of Table II, 0.038, for therelationship between income and urbanization implies that Gua-temala at the time was approximately 42 percent richer thanUruguay (exp (0.038 � 9.2) � 1 � 0.42). According to our estimatein column (1) of Table III, we expect Uruguay today to be 105percent richer than Guatemala (exp (0.078 � 9.2) � 1 � 1.05),which is approximately the current difference in income per cap-ita between these two countries.9
The second column of Table III excludes North African coun-tries for which data quality may be lower. The result is un-
8. Because China was never a formal colony, we do not include it in oursample of ex-colonies. Adding China does not affect our results. For example, withChina, the baseline estimate changes from �0.078 (s.e. � 0.026) to �0.079 (s.e. �0.025). Furthermore, our sample excludes countries that were colonized by Euro-pean powers briefly during the twentieth century, such as Iran, Saudi Arabia, andSyria. If we include these observations, the results are essentially unchanged. Forexample, the baseline estimate changes to �0.072 (s.e. � 0.024).
9. Interestingly, these calculations suggest that not only have relative rank-ings reversed since 1500, but income differences are now much larger than in1500.
1244 QUARTERLY JOURNAL OF ECONOMICS
changed, with a coefficient of �0.101 and standard error of 0.032.Column (3) drops the Americas, which increases both the coeffi-cient and the standard error, but the estimate remains highlysignificant. Column (4) reports the results just for the Americas,where the relationship is somewhat weaker but still significant atthe 8 percent level. Column (5) adds continent dummies to checkwhether the relationship is being driven by differences acrosscontinents. Although continent dummies are jointly significant,the coefficient on urbanization in 1500 is unaffected—it is �0.083with a standard error of 0.030.
One might also be concerned that the relationship is beingdriven mainly by the neo-Europes: United States, Canada, NewZealand, and Australia. These countries are settler colonies builton lands that were inhabited by relatively undeveloped civiliza-tions. Although the contrast between the development experi-ences of these areas and the relatively advanced civilizations ofIndia or Central America is of central importance to the reversal andto our story, one would like to know whether there is anything morethan this contrast in the results of Table III. In column (6) we dropthese observations. The relationship is now weaker, but still nega-tive and statistically significant at the 7 percent level.
In column (7) we control for distance from the equator (theabsolute value of latitude), which does not affect the pattern ofthe reversal—the coefficient on urbanization in 1500 is now�0.072 instead of �0.078 in our baseline specification. Distancefrom the equator is itself insignificant. Column (8), in turn, con-trols for a variety of geography variables that represent the effectof climate, such as measures of temperature, humidity, and soiltype, with little effect on the relationship between urbanization in1500 and income per capita today. The R2 of the regressionincreases substantially, but this simply reflects the addition ofsixteen new variables to this regression (the adjusted R2 in-creases only slightly, to 0.27).
In column (9) we control for a variety of “resources” whichmay have been important for post-1500 development. These in-clude dummies for being an island, for being landlocked, and forhaving coal reserves and a variety of other natural resources (seeAppendix 2 for detailed definitions and sources). Access to the seamay have become more important with the rise of trade, andavailability of coal or other natural resources may have differenteffects at different points in time. Once again, the addition ofthese variables has no effect on the pattern of the reversal.
1245REVERSAL OF FORTUNE
TA
BL
EII
IU
RB
AN
IZA
TIO
NIN
1500
AN
DG
DP
PE
RC
AP
ITA
IN19
95F
OR
FO
RM
ER
EU
RO
PE
AN
CO
LO
NIE
S
Dep
ende
nt
vari
able
islo
gG
DP
per
capi
ta(P
PP
)in
1995
Bas
esa
mpl
e(1
)
Wit
hou
tN
orth
Afr
ica
(2)
Wit
hou
tth
eA
mer
icas
(3)
Just
the
Am
eric
as(4
)
Wit
hco
nti
nen
tdu
mm
ies
(5)
Wit
hou
tn
eo-
Eu
rope
s(6
)
Con
trol
lin
gfo
rla
titu
de(7
)
Con
trol
lin
gfo
rcl
imat
e(8
)
Con
trol
lin
gfo
rre
sou
rces
(9)
Con
trol
lin
gfo
rco
lon
ial
orig
in(1
0)
Con
trol
lin
gfo
rre
ligi
on(1
1)
Urb
aniz
atio
nin
1500
�0.
078
�0.
101
�0.
115
�0.
053
�0.
083
�0.
046
�0.
072
�0.
088
�0.
058
�0.
071
�0.
060
(0.0
26)
(0.0
32)
(0.0
51)
(0.0
29)
(0.0
30)
(0.0
26)
(0.0
25)
(0.0
30)
(0.0
29)
(0.0
28)
(0.0
33)
Asi
adu
mm
y�
1.33
(0.6
1)A
fric
adu
mm
y�
0.53
(0.7
7)A
mer
ica
dum
my
�0.
96(0
.57)
Lat
itu
de1.
42(0
.92)
P-v
alu
efo
rte
mpe
ratu
re[0
.51]
P-v
alu
efo
rh
um
idit
y[0
.40]
P-v
alu
efo
rso
ilqu
alit
y[0
.96]
P-v
alu
efo
rre
sou
rces
[0.1
6]
1246 QUARTERLY JOURNAL OF ECONOMICS
Lan
dloc
ked
�0.
54(0
.48)
Isla
nd
0.27
(0.3
3)C
oal
0.11
(0.2
8)F
orm
erF
ren
chco
lon
y�
0.59
(0.3
9)F
orm
erS
pan
ish
colo
ny
0.06
(0.2
9)P
-val
ue
for
reli
gion
[0.4
7]
R2
0.19
0.22
0.26
0.13
0.32
0.09
0.24
0.53
0.45
0.27
0.25
Nu
mbe
rof
obse
rvat
ion
s41
3717
2441
3741
4141
4141
Sta
nda
rder
rors
are
inpa
ren
thes
es.
P-v
alu
esfr
omF
-tes
tsfo
rjo
int
sign
ifica
nce
are
insq
uar
ebr
acke
ts.
Dep
ende
nt
vari
able
islo
gG
DP
per
capi
ta(P
PP
)in
1995
.B
ase
sam
ple
isal
lfor
mer
colo
nie
sfo
rw
hic
hw
eh
ave
data
.Urb
aniz
atio
nin
1500
ispe
rcen
tof
the
popu
lati
onli
vin
gin
tow
ns
wit
h50
00or
mor
ein
hab
itan
ts.T
he
regr
essi
onth
atin
clu
des
con
tin
ent
dum
mie
sh
asO
cean
iaas
the
base
cate
gory
.T
he
neo
-Eu
rope
sar
eth
eU
nit
edS
tate
s,C
anad
a,A
ust
rali
a,an
dN
ewZ
eala
nd.
Inth
e“c
lim
ate”
regr
essi
onw
ein
clu
defi
vem
easu
res
ofte
mpe
ratu
re,
fou
rm
easu
res
ofh
um
idit
y,an
dse
ven
mea
sure
sof
soil
qual
ity.
Inth
e“r
esou
rces
”re
gres
sion
we
incl
ude
rese
rves
ofgo
ld,i
ron
,zin
c,si
lver
,an
doi
l.C
oali
sa
dum
my
for
the
pres
ence
ofco
al,l
andl
ocke
dis
adu
mm
yfo
rn
oth
avin
gac
cess
toth
ese
a,an
dis
lan
dis
adu
mm
yfo
rbe
ing
anis
lan
d.T
he
regr
essi
onth
atco
ntr
ols
for
colo
nia
lor
igin
incl
ude
sdu
mm
ies
for
form
erF
ren
chco
lon
y,S
pan
ish
colo
ny,
Por
tugu
ese
colo
ny,
Bel
gian
colo
ny,
Ital
ian
colo
ny,
Ger
man
colo
ny,
and
Du
tch
colo
ny.
Bri
tish
colo
nie
sar
eth
eba
seca
tego
ry.
Th
ere
ligi
onva
riab
les
are
perc
ent
ofth
epo
pula
tion
wh
oar
eM
usl
im,
Cat
hol
ic,
and
“oth
er”;
perc
ent
Pro
test
ant
isth
eba
seca
tego
ry.
For
deta
iled
sou
rces
and
desc
ript
ion
sse
eA
ppen
dix
2.
1247REVERSAL OF FORTUNE
Finally, in columns (10) and (11) we add the identity of thecolonial power and religion, which also have little effect on ourestimate, and are themselves insignificant.
The urbanization variable used in Table III relies on work byBairoch and Eggimann. In Table IV we use data from Bairoch andEggimann separately, as well as data from Chandler, who pro-vided the starting point for Bairoch’s data. We report a subset ofthe regressions from Table III using these three different seriesand an alternative series using the Davis-Zipf adjustment toconvert Eggimann’s estimates into Bairoch-equivalent numbers(explained in Appendix 1). The results are very similar to thebaseline estimates reported in Table III: in all cases, there is anegative relationship between urbanization in 1500 and incomeper capita today, and in almost all cases, this relationship isstatistically significant at the 5 percent level (the full set ofresults are reported in Acemoglu, Johnson, and Robinson [2001b]).
III.B. Results with Population Density
In Panel A of Table V we regress income per capita today onlog population density in 1500, and also include data for sub-Saharan Africa. The results are similar to those in Table IV (alsosee Figure II). In all specifications we find that countries withhigher population density in 1500 are substantially poorer today.The coefficient of �0.38 in column (1) implies that a 10 percenthigher population density in 1500 is associated with a 4 percentlower income per capita today. For example, the area now corre-sponding to Bolivia was seven times more densely settled thanthe area corresponding to Argentina; so on the basis of thisregression, we expect Argentina to be three times as rich asBolivia, which is more or less the current gap in income betweenthese countries.10
The remaining columns perform robustness checks, andshow that including a variety of controls for geography and re-sources, the identity of the colonial power, religion variables, ordropping the Americas, the neo-Europes, or North Africa has very
10. The magnitudes implied by the estimates in this table are similar to thoseimplied by the estimates in Table III. For example, the difference in the urban-ization rate between an average high and low urbanization country in 1500 is 8.1(see columns (4) and (5) in Table I), which using the coefficient of �0.078 fromTable III translates into a 0.078 � 8.1 � 0.63 log points difference in current GDP.The difference in log population density between an average high-density andlow-density country in 1500 is 2.2 (see columns (6) and (7) in Table I), whichtranslates into a 0.38 � 2.2 � 0.84 log points difference in current GDP.
1248 QUARTERLY JOURNAL OF ECONOMICS
little effect on the results. In all cases, log population density in1500 is significant at the 1 percent level (although now some ofthe controls, such as the humidity dummies, are also significant).
TABLE IVALTERNATIVE MEASURES OF URBANIZATION
Dependent variable is log GDP per capita (PPP) in 1995
Basesample
(1)
With continentdummies
(2)
Withoutneo-Europes
(3)
Controllingfor latitude
(4)
Controllingfor resources
(5)
Panel A: Using our base sample measure of urbanization
Urbanization in 1500 �0.078 �0.083 �0.046 �0.072 �0.058(0.026) (0.030) (0.026) (0.025) (0.029)
R2 0.19 0.32 0.09 0.24 0.45Number of observations 41 41 37 41 41
Panel B: Using only Bairoch’s estimates
Urbanization in 1500 �0.126 �0.107 �0.089 �0.116 �0.092(0.032) (0.034) (0.033) (0.036) (0.037)
R2 0.30 0.37 0.19 0.31 0.49Number of observations 37 37 33 37 37
Panel C: Using only Eggimann’s estimates
Urbanization in 1500 �0.041 �0.043 �0.022 �0.036 �0.022(0.019) (0.019) (0.018) (0.019) (0.023)
R2 0.10 0.28 0.04 0.16 0.39Number of observations 41 41 37 41 41
Panel D: Using only Chandler’s estimates
Urbanization in 1500 �0.057 �0.072 �0.040 �0.054 �0.049(0.019) (0.021) (0.019) (0.019) (0.025)
R2 0.27 0.43 0.17 0.34 0.66Number of observations 26 26 23 26 26
Panel E: Using Davis-Zipf Adjustment for Eggimann’s series
Urbanization in 1500 �0.039 �0.048 �0.024 �0.040 �0.031(0.015) (0.020) (0.014) (0.015) (0.017)
R2 0.14 0.30 0.08 0.23 0.44Number of observations 41 41 37 41 41
Standard errors are in parentheses. Dependent variable is log GDP per capita (PPP) in 1995. Basesample is all former colonies for which we have data. Urbanization in 1500 is percent of the population livingin towns with 5000 or more people. In Panels B, C, D, and E, we use, respectively, Bairoch’s estimates,Eggimann’s estimates, Chandler’s estimates, and a conversion of Eggimann’s estimates into Bairoch-equiva-lent numbers using the Davis-Zipf adjustment. Eggimann’s estimates (Panel C) and Chandler’s estimates(Panel D) are not converted to Bairoch-equivalent units. The continent dummies, neo-Europes, and resourcesmeasures are as described in the note to Table III. For detailed sources and descriptions see Appendix 2. Thealternative urbanization series are shown in Appendix 3.
1249REVERSAL OF FORTUNE
TA
BL
EV
PO
PU
LA
TIO
ND
EN
SIT
YA
ND
GD
PP
ER
CA
PIT
AIN
FO
RM
ER
EU
RO
PE
AN
CO
LO
NIE
S
Dep
ende
nt
vari
able
islo
gG
DP
per
capi
ta(P
PP
)in
1995
Bas
esa
mpl
e(1
)
Wit
hou
tA
fric
a(2
)
Wit
hou
tth
eA
mer
icas
(3)
Just
the
Am
eric
as(4
)
Wit
hco
nti
nen
tdu
mm
ies
(5)
Wit
hou
tn
eo-
Eu
rope
s(6
)
Con
trol
lin
gfo
rla
titu
de(7
)
Con
trol
lin
gfo
rcl
imat
e(8
)
Con
trol
lin
gfo
rre
sou
rces
(9)
Con
trol
lin
gfo
rco
lon
ial
orig
in(1
0)
Con
trol
lin
gfo
rre
ligi
on(1
1)
Pan
elA
:L
ogpo
pula
tion
den
sity
in15
00as
ind
epen
den
tva
riab
le
Log
popu
lati
onde
nsi
tyin
1500
�0.
38�
0.40
�0.
32�
0.25
�0.
26�
0.32
�0.
33�
0.31
�0.
30�
0.32
�0.
37(0
.06)
(0.0
5)(0
.07)
(0.0
9)(0
.05)
(0.0
6)(0
.06)
(0.0
6)(0
.06)
(0.0
6)(0
.07)
Asi
adu
mm
y�
0.91
(0.5
5)A
fric
adu
mm
y�
1.67
(0.5
2)A
mer
ica
dum
my
�0.
69(0
.51)
Lat
itu
de2.
09(0
.74)
P-v
alu
efo
rte
mpe
ratu
re[0
.18]
P-v
alu
efo
rh
um
idit
y[0
.00]
P-v
alu
efo
rso
ilqu
alit
y[0
.10]
P-v
alu
efo
rn
atu
ral
reso
urc
es[0
.34]
Lan
dloc
ked
�0.
58(0
.23)
Isla
nd
0.62
(0.2
3)
1250 QUARTERLY JOURNAL OF ECONOMICS
Coa
l0.
01(0
.19)
For
mer
Fre
nch
colo
ny
�0.
48(0
.20)
For
mer
Spa
nis
hco
lon
y0.
25(0
.22)
P-v
alu
efo
rre
ligi
on[0
.73]
R2
0.34
0.55
0.27
0.22
0.56
0.24
0.40
0.59
0.54
0.48
0.36
Nu
mbe
rof
obse
rvat
ion
s91
4758
3391
8791
9085
9185
Pan
elB
:L
ogpo
pula
tion
and
log
lan
din
1500
asse
para
tein
dep
end
ent
vari
able
s
Log
popu
lati
onin
1500
�0.
34�
0.30
�0.
32�
0.13
�0.
23�
0.27
�0.
29�
0.27
�0.
27�
0.28
�0.
31(0
.05)
(0.0
5)(0
.07)
(0.0
7)(0
.05)
(0.0
5)(0
.05)
(0.0
5)(0
.05)
(0.0
5)(0
.06)
Log
arab
lela
nd
in15
000.
260.
270.
210.
160.
180.
150.
200.
200.
080.
210.
24(0
.06)
(0.0
6)(0
.09)
(0.0
6)(0
.05)
(0.0
6)(0
.06)
(0.0
6)(0
.07)
(0.0
6)(0
.07)
R2
0.35
0.45
0.31
0.17
0.55
0.31
0.41
0.59
0.55
0.47
0.36
Nu
mbe
rof
obse
rvat
ion
s91
4758
3391
8791
9085
9185
Pan
elC
:U
sin
gpo
pula
tion
den
sity
in10
00A
.D.
asan
inst
rum
ent
for
popu
lati
ond
ensi
tyin
1500
A.D
.
Log
popu
lati
onde
nsi
tyin
1500
�0.
31�
0.4
�0.
15�
0.38
�0.
18�
0.22
�0.
27�
0.26
�0.
22�
0.26
�0.
25(0
.06)
(0.0
6)(0
.08)
(0.1
1)(0
.07)
(0.0
8)(0
.06)
(0.0
7)(0
.07)
(0.0
6)(0
.08)
Nu
mbe
rof
obse
rvat
ion
s83
4351
3283
8083
8378
8377
Sta
nda
rder
rors
are
inpa
ren
thes
es.
P-v
alu
esfr
omF
-tes
tsfo
rjo
int
sign
ifica
nce
are
insq
uar
ebr
acke
ts.
Dep
ende
nt
vari
able
islo
gG
DP
per
capi
ta(P
PP
)in
1995
.B
ase
sam
ple
isal
lfor
mer
colo
nie
sfo
rw
hic
hw
eh
ave
data
.Pop
ula
tion
den
sity
in15
00is
tota
lpop
ula
tion
divi
ded
byar
able
lan
dar
ea.S
eeT
able
III
for
anex
plan
atio
nof
the
sam
ple
and
cova
riat
esin
each
colu
mn
.F
orde
tail
edso
urc
esan
dde
scri
ptio
ns
see
App
endi
x2.
1251REVERSAL OF FORTUNE
The estimates in the top panel of Table V use variation inpopulation density, which reflects two components: differences inpopulation and differences in arable land area. In Panel B weseparate the effects of these two components and find that theycome in with equal and opposite signs, showing that the specifi-cation with population density is appropriate. In Panel C we usepopulation density in 1000 as an instrument for population den-sity in 1500. This is useful since, as discussed in subsection II.C,differences in long-run population density are likely to be betterproxies for income per capita. Instrumenting for population den-sity in 1500 with population density in 1000 isolates the long-runcomponent of population density differences across countries (i.e.,the component of population density in 1500 that is correlatedwith population density in 1000). The Two-Stage Least Squares(2SLS) results in Panel C using this instrumental variables strat-egy are very similar to the OLS results in Panel A.
III.C. Further Results, Robustness Checks, and Discussion
Caution is required in interpreting the results presented inTables III, IV, and V. Estimates of urbanization and population in1500 are likely to be error-ridden. Nevertheless, the first effect ofmeasurement error would be to create an attenuation bias toward0. Therefore, one might think that the negative coefficients inTables III, IV, and V are, if anything, underestimates. A moreserious problem would be if errors in the urbanization and popu-lation density estimates were not random, but correlated withcurrent income in some systematic way. We investigate this issuefurther in Table VI, using a variety of different estimates forurbanization and population density. Columns (1)–(5), for exam-ple, show that the results are robust to a variety of modificationsto the urbanization data.
Much of the variation in urbanization and population densityin 1500 was not at the level of these countries, but at the level of“civilizations.” For example, in 1500 there were fewer separatecivilizations in the Americas, and even arguably in Asia, thanthere are countries today. For this reason, in column (6) we repeatour key regressions using variation in urbanization and popula-tion density only among fourteen civilizations (based on Toynbee[1934–1961] and McNeill [1999]—see the note to Table VI). Theresults confirm our basic findings, and show a statistically signifi-cant negative relationship between prosperity in 1500 and today.Columns (7) and (8) report robustness checks using variants of
1252 QUARTERLY JOURNAL OF ECONOMICS
the population density data constructed under different assump-tions, again with very similar results.
Is there a similar reversal among the noncolonies? Column(9) reports a regression of log GDP per capita in 1995 on urban-ization in 1500 for all noncolonies (including Europe), and column(10) reports the same regression for Europe (including EasternEurope). In both cases, there is a positive relationship betweenurbanization in 1500 and income today.11 This suggests that thereversal reflects an unusual event, and is likely to be related tothe effect of European colonialism on these societies.
Panel B of Table VI reports results weighted by population in1500, with very similar results. In Panel C we include urbaniza-tion and population density simultaneously in these regressions.In all cases, population density is negative and highly significant,while urbanization is insignificant. This is consistent with thenotion, discussed below, that differences in population densityplayed a key role in the reversal in relative incomes among thecolonies (although it may also reflect measurement error in theurbanization estimates).
As a final strategy to deal with the measurement error inurbanization, we use log population density as an instrument forurbanization rates in 1500. When both of these are valid proxiesfor economic prosperity in 1500 and the measurement error isclassical, this procedure corrects for the measurement error prob-lem. Not surprisingly, these instrumental-variables estimatesreported in the bottom panel of Table VI are considerably largerthan the OLS estimates in Table III. For example, the baselineestimate is now �0.18 instead of �0.08 in Table III. The generalpattern of reversal in relative incomes is unchanged, however.
Is the reversal shown in Figures I and II and Tables III, IV,and V consistent with other evidence? The literature on thehistory of civilizations documents that 500 years ago many partsof Asia were highly prosperous (perhaps as prosperous as West-ern Europe), and civilizations in Meso-America and North Africawere relatively developed (see, e.g., Abu-Lughod [1989], Braudel[1992], Chaudhuri [1990], Hodgson [1993], McNeill [1999], Po-meranz [2000], Reid [1988, 1993], and Townsend [2000]). In con-
11. In Acemoglu, Johnson, and Robinson [2001b] we also provided evidencethat urbanization and population density in 1000 are positively correlated withurbanization and population density in 1500, suggesting that before 1500 therewas considerable persistence in prosperity both where the Europeans later colo-nized and where they never colonized.
1253REVERSAL OF FORTUNE
TA
BL
EV
IR
OB
US
TN
ES
SC
HE
CK
SF
OR
UR
BA
NIZ
AT
ION
AN
DL
OG
PO
PU
LA
TIO
ND
EN
SIT
Y
Dep
ende
nt
vari
able
islo
gG
DP
per
capi
ta(P
PP
)in
1995
Bas
esa
mpl
e(1
)
Ass
um
ing
low
eru
rban
izat
ion
inth
eA
mer
icas
(2)
Ass
um
ing
low
eru
rban
izat
ion
inN
orth
Afr
ica
(3)
Ass
um
ing
low
eru
rban
izat
ion
inIn
dian
subc
onti
nen
t(4
)
Usi
ng
leas
tfa
vora
ble
com
bin
atio
nof
assu
mpt
ion
s(5
)
Usi
ng
augm
ente
dT
oyn
bee
defi
nit
ion
ofci
vili
zati
on(6
)
Usi
ng
lan
dar
eain
1995
for
popu
lati
onde
nsi
ty(7
)
Alt
ern
ativ
eas
sum
ptio
ns
for
log
popu
lati
onde
nsi
ty(8
)
All
cou
ntr
ies
nev
erco
lon
ized
byE
uro
pe(9
)
Eu
rope
(in
clu
din
gE
aste
rnE
uro
pe)
(10)
For
mer
colo
nie
sN
ever
colo
niz
ed
Pan
elA
:U
nw
eigh
ted
regr
essi
ons
Urb
aniz
atio
nin
1500
�0.
078
�0.
089
�0.
102
�0.
073
�0.
105
�0.
117
0.06
80.
077
(0.0
26)
(0.0
27)
(0.0
29)
(0.0
27)
(0.0
32)
(0.0
52)
(0.0
23)
(0.0
23)
Log
popu
lati
onde
nsi
tyin
1500
�0.
41�
0.32
(0.0
6)(0
.07)
R2
0.20
0.22
0.24
0.16
0.21
0.30
0.35
0.21
0.18
0.27
Nu
mbe
rof
obse
rvat
ion
s41
4141
4141
1491
9143
32
Pan
elB
:R
egre
ssio
ns
wei
ghte
du
sin
glo
gpo
pula
tion
in15
00
Urb
aniz
atio
nin
1500
�0.
072
�0.
084
�0.
097
�0.
064
�0.
099
�0.
118
�0.
064
�0.
073
(0.0
25)
(0.0
26)
(0.0
29)
(0.0
26)
(0.0
32)
(0.0
53)
(0.0
23)
(0.0
22)
Log
popu
lati
onde
nsi
tyin
1500
�0.
39�
0.29
(0.0
6)(0
.07)
R2
0.18
0.22
0.23
0.14
0.20
0.29
0.32
0.19
0.17
0.24
Nu
mbe
rof
obse
rvat
ion
s41
4141
4141
1491
9143
32
1254 QUARTERLY JOURNAL OF ECONOMICS
Pan
elC
:In
clu
din
gbo
thu
rban
izat
ion
and
log
popu
lati
ond
ensi
tyas
ind
epen
den
tva
riab
les
Urb
aniz
atio
nin
1500
0.03
80.
039
0.01
70.
037
0.02
00.
072
0.01
70.
003
0.02
80.
032
(0.0
28)
(0.0
31)
(0.0
33)
(0.0
27)
(0.0
35)
(0.0
47)
(0.0
23)
(0.0
22)
(0.0
20)
(0.0
21)
Log
popu
lati
onde
nsi
tyin
1500
�0.
41�
0.41
�0.
36�
0.40
�0.
37�
0.48
�0.
43�
0.41
0.34
0.37
(0.0
7)(0
.08)
(0.0
7)(0
.07)
(0.0
7)(0
.09)
(0.0
7)(0
.07)
(0.0
7)(0
.08)
R2
0.56
0.56
0.54
0.56
0.54
0.79
0.61
0.60
0.48
0.57
Nu
mbe
rof
obse
rvat
ion
s41
4141
4141
1441
4143
32
Pan
elD
:In
stru
men
tin
gfo
ru
rban
izat
ion
in15
00u
sin
glo
gpo
pula
tion
den
sity
in15
00
Urb
aniz
atio
nin
1500
�0.
178
�0.
181
�0.
215
�0.
194
�0.
242
�0.
237
�0.
217
�0.
239
0.25
90.
226
(0.0
4)(0
.040
)(0
.048
)(0
.048
)(0
.057
)(0
.080
)(0
.053
)(0
.063
)(0
.090
)(0
.074
)N
um
ber
ofob
serv
atio
ns
4141
4141
4114
4141
4332
Sta
nda
rder
rors
are
inpa
ren
thes
es.
Dep
ende
nt
vari
able
islo
gG
DP
per
capi
ta(P
PP
)in
1995
.B
ase
sam
ple
isal
lfo
rmer
colo
nie
sfo
rw
hic
hw
eh
ave
data
.In
our
base
sam
ple,
urb
aniz
atio
nin
1500
ispe
rcen
tof
the
popu
lati
onli
vin
gin
tow
ns
wit
h50
00or
mor
epe
ople
.Col
um
n(2
)as
sum
es9
perc
ent
urb
aniz
atio
nin
the
An
des
and
Cen
tral
Am
eric
a.C
olu
mn
(3)
assu
mes
10pe
rcen
tu
rban
izat
ion
inN
orth
Afr
ica.
Col
um
n(4
)as
sum
es6
perc
ent
urb
aniz
atio
nin
the
Indi
ansu
bcon
tin
ent.
Col
um
n(5
)co
mbi
nes
the
assu
mpt
ion
sof
colu
mn
s(2
),(3
),(4
),an
d(5
)to
crea
teth
ele
ast
favo
rabl
eco
mbi
nat
ion
ofas
sum
ptio
ns
for
our
hyp
oth
esis
.C
olu
mn
(6)
ison
lyci
vili
zati
ons
info
rmer
Eu
rope
anco
lon
ies.
Th
eau
gmen
ted
Toy
nbe
eci
vili
zati
ons,
use
din
colu
mn
(6),
incl
ude
An
dean
,M
exic
,Y
uca
tec,
Ara
bic
(Nor
thA
fric
a),
Hin
du,
Pol
ynes
ian
,E
skim
o(C
anad
a)N
orth
Am
eric
anIn
dian
,S
outh
Am
eric
anIn
dian
(Bra
zil/A
rgen
tin
a/C
hil
e),
Au
stra
lian
Abo
rigi
ne,
Mal
ay(M
alay
sia
and
Indo
nes
ia),
Ph
ilip
pin
es,
Vie
tnam
/Cam
bodi
a,an
dB
urm
a.In
colu
mn
(7)
popu
lati
onde
nsi
tyin
1500
isto
tal
popu
lati
ondi
vide
dby
arab
lela
nd
area
in19
95.
Col
um
n(8
)h
alve
sth
epo
pula
tion
den
sity
esti
mat
esfo
rA
fric
a.F
orde
tail
edso
urc
esan
dde
scri
ptio
ns
see
App
endi
x2.
1255REVERSAL OF FORTUNE
trast, there was little agriculture in most of North America andAustralia, at most consistent with a population density of 0.1people per square kilometer. McEvedy and Jones [1978, p. 322]describe the state of Australia at this time as “an unchangingpalaeolithic backwater.” In fact, because of the relative backward-ness of these areas, European powers did not view them asvaluable colonies. Voltaire is often quoted as referring to Canadaas a “few acres of snow,” and the European powers at the timepaid little attention to Canada relative to the colonies in the WestIndies. In a few parts of North America, along the East Coast andin the Southwest, there was settled agriculture, supporting apopulation density of approximately 0.4 people per square kilo-meter, but this was certainly much less than that in the Aztec andInca Empires, which had fully developed agriculture with a popu-lation density of between 1 and 3 people (or even higher) persquare kilometer, and also much less than the correspondingnumbers in Asia and Africa [McEvedy and Jones 1978, p. 273].The recent work by Maddison [2001] also confirms our interpre-tation. He estimates that India, Indonesia, Brazil, and Mexicowere richer than the United States in 1500 and 1700 (see, forexample, his Table 2-22a).
III.D. The Timing and Nature of the Reversal
The evidence presented so far documents the reversal inrelative incomes among the former colonies from 1500 to today.When did this reversal take place? This question is relevant inthinking about the causes of the reversal. For example, if thereversal is related to the extraction of resources from, and the“plunder” of, the former colonies, or to the direct effect of thediseases Europeans brought to the New World, it should havetaken place shortly after colonization.
Figure IV shows that the reversal is mostly a late eighteenth-and early nineteenth-century phenomenon, and is closely relatedto industrialization. Figure IVa compares the evolution of urban-ization among two groups of New World ex-colonies, those withlow urbanization in 1500 versus those with high urbanization in1500.12 We focus on New World colonies since the societies came
12. The initially high urbanization countries for which we have data and areincluded in the figure are Bolivia, Mexico, Peru, and all of Central America, whilethe initially low urbanization countries are Argentina, Brazil, Canada, Chile, andthe United States.
1256 QUARTERLY JOURNAL OF ECONOMICS
under European dominance very early on. The averages plottedin the figure are weighted by population in 1500. In addition, inthe same figure we plot India and the United States separately(as well as including it in the initially low urbanization group).The figure shows that the initially low urbanization group as awhole and the United States by itself overtake India and the ini-tially high urbanization countries sometime between 1750 and 1850.
Figure IVb depicts per capita industrial production for theUnited States, Canada, New Zealand, Australia, Brazil, Mexico,and India using data from Bairoch [1982]. This figure shows thetakeoff in industrial production in the United States, Australia,Canada, and New Zealand relative to Brazil, Mexico, and India.Although the scale makes it difficult to see in the figure, percapita industrial production in 1750 was in fact higher in India, 7,than in the United States, 4 (with U. K. industrial production percapita in 1900 normalized to 100). Bairoch [1982] also reportsthat in 1750 China had industrial production per capita twice the
FIGURE IVaUrbanization Rate in India, the United States, and New World Countries
with Low and High Urbanization, 800–1920Note. Urbanization is population living in urban areas divided by total popula-
tion. Urban areas have a minimum threshold of 20,000 inhabitants, from Chan-dler [1987], and Mitchell [1993, 1995]. Low urbanization in 1500 countries areArgentina, Brazil, Canada, Chile, and the United States. High urbanization in1500 countries are Bolivia, Ecuador, Mexico, Peru, and all of Central America. Fordetails see Appendix 1.
1257REVERSAL OF FORTUNE
level of the United States. Yet, as Figure IVb shows, over the next200 years there was a much larger increase in industrial produc-tion in the United States than in India (and also than in China).
This general interpretation, that the reversal in relative in-comes took place during the late eighteenth and early nineteenthcenturies and was linked to industrialization, is also consistent withthe fragmentary evidence we have on other measures of income percapita and industrialization. Coatsworth [1993], Eltis [1995], Enger-man [1981], and Engerman and Sokoloff [1997] provide evidencethat much of Spanish America and the Caribbean were more pros-perous (had higher per capita income) than British North Americauntil the eighteenth century. The future United States rose in percapita income during the 1700s relative to the Caribbean and SouthAmerica, but only really pulled ahead during the late eighteenthand early nineteenth centuries. Maddison’s [2001] numbers alsoshow that India, Indonesia, Brazil, and Mexico were richer than theUnited States in 1700, but had fallen behind by 1820.
U. S. growth during this period also appears to be an indus-try-based phenomenon. McCusker and Menard [1985] and Galen-son [1996] both emphasize that productivity and income growthin North America before the eighteenth century was limited.During the critical period of growth in the United States, between
FIGURE IVbIndustrial Production per Capita, 1750–1953
Note. Index of industrial production with U. K. per capita industrialization in1900 is equal to 100, from Bairoch [1982].
1258 QUARTERLY JOURNAL OF ECONOMICS
1840 and 1900, there was modest growth in agricultural outputper capita, and very rapid growth in industrial output per capita;the numbers reported by Gallman [2000] imply that between1840 and 1900 agricultural product per capita increased by about30 percent, a very small increase relative to the growth in manu-facturing output per capita, which increased more than fourfold.
IV. HYPOTHESES AND EXPLANATIONS
IV.A. The Geography Hypothesis
The geography hypothesis claims that differences in eco-nomic performance reflect differences in geographic, climatic, andecological characteristics across countries. There are many differ-ent versions of this hypothesis. Perhaps the most common is theview that climate has a direct effect on income through its influ-ence on work effort. This idea dates back to Machiavelli [1519]and Montesquieu [1748]. Both Toynbee [1934, Vol. 1] and Mar-shall [1890, p. 195] similarly emphasized the importance of cli-mate, both on work effort and productivity. One of the pioneers ofdevelopment economics, Myrdal [1968], also placed considerableemphasis on the effect of geography on agricultural productivity.He argued: “serious study of the problems of underdevelop-ment . . . should take into account the climate and its impacts onsoil, vegetation, animals, humans and physical assets—in short,on living conditions in economic development” [Vol. 3, p. 2121].
More recently, Diamond [1997] and Sachs [2000, 2001] haveespoused different versions of the geography view. Diamond, forexample, argues that the timing of the Neolithic revolution hashad a long-lasting effect on economic and social development.Sachs, on the other hand, emphasizes the importance of geogra-phy through its effect on the disease environment, transportcosts, and technology. He writes: “Certain parts of the world aregeographically favored. Geographical advantages might includeaccess to key natural resources, access to the coastline and sea—navigable rivers, proximity to other successful economies, advan-tageous conditions for agriculture, advantageous conditions forhuman health” [2000, p. 30]. Also see Myrdal [1968, Vol. 1, pp.691–695].
This simple version of the geography hypothesis predictspersistence in economic outcomes, since the geographic factorsthat are the first-order determinants of prosperity are time-in-
1259REVERSAL OF FORTUNE
variant. The evidence presented so far therefore weighs againstthe simple geography hypothesis: whatever factors are importantin making former colonies rich today are very different from thosecontributing to prosperity in 1500.
IV.B. The Sophisticated Geography Hypotheses
The reversal in relative incomes does not necessarily reject amore sophisticated geography hypothesis, however. Certain geo-graphic characteristics that were not useful, or that were evenharmful, for successful economic performance in 1500 may turnout to be beneficial later on. In this subsection we briefly discussa number of sophisticated geography hypotheses emphasizing theimportance of such time-varying effects of geography.13
The first is the “temperate drift hypothesis,” emphasizing thetemperate (or away from the equator) shift in the center of eco-nomic gravity over time. According to this view, geography be-comes important when it interacts with the presence of certaintechnologies. For example, one can argue that tropical areasprovided the best environment for early civilizations—after all,humans evolved in the tropics, and the required calorie intake islower in warmer areas. But with the arrival of “appropriate”technologies, temperate areas became more productive. The tech-nologies that were crucial for progress in temperate areas includethe heavy plow, systems of crop rotation, domesticated animalssuch as cattle and sheep, and some of the high productivityEuropean crops, including wheat and barley. Despite the key roleof these technologies for temperate areas, they have had muchless of an effect on tropical zones [Lewis 1978]. Sachs [2001, p. 12]also implies this view in his recent paper when he adapts Dia-mond’s argument about the geography of technological diffusion:“Since technologies in the critical areas of agriculture, health, andrelated areas could diffuse within ecological zones, but not acrossecological zones, economic development spread through the tem-
13. Put differently, in the simple geography hypothesis, geography has a main effecton economic performance, which can be expressed as Yit � �0 � �1 � Gi � �t � �it,where Yit is a measure of economic performance in country i at time t, Gi is a measureof geographic characteristics, �t is a time effect, and �it measures other country-time-specific factors. In contrast, in the sophisticated geography view, the relationshipbetween income and geography would be Yit � �0 � �1 � Git � �2 � Tt � Git � �t � �it,where Tt is a time-varying characteristic of the world as a whole or of the state oftechnology. According to this view, the major role that geography plays in history isnot through �1, but through �2.
1260 QUARTERLY JOURNAL OF ECONOMICS
perate zones but not through the tropical regions” (italics in theoriginal; also see Myrdal [1968], Ch. 14).
The evidence is not favorable to the view that the reversalreflects the emergence of agricultural technologies favorable totemperate areas, however. First, the regressions in Tables III, IV,and V show little evidence that the reversal was related to geo-graphic characteristics. Second, the temperate drift hypothesissuggests that the reversal should be associated with the spread ofEuropean agricultural technologies. Yet in practice, while Euro-pean agricultural technology spread to the colonies between thesixteenth and eighteenth centuries (e.g., McCusker and Menard[1985], Ch. 3 for North America), the reversal in relative incomesis largely a late eighteenth- and early nineteenth-century, andindustry-based phenomenon.
In light of the result that the reversal is related to industri-alization, another sophisticated geography hypothesis would bethat certain geographic characteristics facilitate or enable indus-trialization. First, one can imagine that there is more room forspecialization in industry, but such specialization requires trade.If countries differ according to their transport costs, it might bethose with low transport costs that take off during the age ofindustry. This argument is not entirely convincing, however,again because there is little evidence that the reversal was re-lated to geographic characteristics (see Tables III, IV, and V).Moreover, many of the previously prosperous colonies that failedto industrialize include islands such as the Caribbean, or coun-tries with natural ports such as those in Central America, India,or Indonesia. Moreover, transport costs appear to have beenrelatively low in some of the areas that failed to industrialize(e.g., Pomeranz [2000], Appendix A).
Second, countries may lack certain resource endowments,most notably coal, which may have been necessary for industri-alization (e.g., Pomeranz [2000] and Wrigley [1988]). But coal isone of the world’s most common resources, with proven reservesin 100 countries and production in over 50 countries [World CoalInstitute 2000], and our results in Table III and V offer littleevidence that either coal or the absence of any other resource wasresponsible for the reversal. So there appears to be little supportfor these types of sophisticated geography hypotheses either.14
14. Two other related hypotheses are worth mentioning. First, it could beargued that people work less hard in warmer climates and that this matters more
1261REVERSAL OF FORTUNE
IV.C. The Institutions Hypothesis
According to the institutions hypothesis, societies with asocial organization that provides encouragement for investmentwill prosper. Locke [1980], Smith [1778], and Hayek [1960],among many others, emphasized the importance of propertyrights for the success of nations. More recently, economists andhistorians have emphasized the importance of institutions thatguarantee property rights. For example, Douglass North startshis 1990 book by stating [p. 3]: “That institutions affect theperformance of economies is hardly controversial,” and identifieseffective protection of property rights as important for the orga-nization of society (see also North and Thomas [1973] and Olson[2000]).
In this context we take a good organization of society tocorrespond to a cluster of (political, economic, and social) institu-tions ensuring that a broad cross section of society has effectiveproperty rights. We refer to this cluster as institutions of privateproperty, and contrast them with extractive institutions, wherethe majority of the population faces a high risk of expropriationand holdup by the government, the ruling elite, or other agents.Two requirements are implicit in this definition of institutions ofprivate property. First, institutions should provide secure prop-erty rights, so that those with productive opportunities expect toreceive returns from their investments, and are encouraged toundertake such investments. The second requirement is embed-ded in the emphasis on “a broad cross section of the society.” Asociety in which a very small fraction of the population, forexample, a class of landowners, holds all the wealth and politicalpower may not be the ideal environment for investment, even if
for industry than for agriculture, thus explaining the reversal. However, there isno evidence either for the hypothesis that work effort matters more for industryor for the assertion that human energy output depends systematically on tempera-ture (see, e.g., Collins and Roberts [1988]). Moreover, the available evidence onhours worked indicates that people work harder in poorer/warmer countries (e.g.,ILO [1995, pp. 36–37]), though of course these high working hours could reflectother factors.
Second, it can be argued that different paths of development reflect the directinfluence of Europeans. Places where there are more Europeans have becomericher, either because Europeans brought certain values conducive to develop-ment (e.g., Landes [1998], and Hall and Jones [1999]), or because having moreEuropeans confers certain benefits (e.g., through trade with Europe or becauseEuropeans are more productive). In Acemoglu, Johnson, and Robinson [2001b] wepresented evidence showing that the reversal and current income levels are notrelated to the current racial composition of the population or to proxies of whetherthe colonies were culturally or politically dominated by Europeans.
1262 QUARTERLY JOURNAL OF ECONOMICS
the property rights of this elite are secure. In such a society, manyof the agents with the entrepreneurial human capital and invest-ment opportunities may be those without effective property rightsprotection. In particular, the concentration of political and socialpower in the hands of a small elite implies that the majority of thepopulation risks being held up by the powerful elite after theyundertake investments. This is also consistent with North andWeingast’s [1989, pp. 805–806] emphasis that what matters is:“ . . . whether the state produces rules and regulations that bene-fit a small elite and so provide little prospect for long-run growth,or whether it produces rules that foster long-term growth.”Whether political power is broad-based or concentrated in thehands of a small elite is crucial in evaluating the role of institu-tions in the experiences of the Caribbean or India during colonialtimes, where the property rights of the elite were well enforced,but the majority of the population had no civil rights or propertyrights.
It is important to emphasize that “equilibrium institutions”may be extractive, even though such institutions do not encour-age economic development. This is because institutions areshaped, at least in part, by politically powerful groups that mayobtain fewer rents with institutions of private property (e.g.,North [1990]), or fear losing their political power if there isinstitutional development (e.g., Acemoglu and Robinson [2000,2001]), or simply may be reluctant to initiate institutional changebecause they would not be the direct beneficiaries of the resultingeconomic gains. In the context of the development experience ofthe former colonies, this implies that equilibrium institutions arelikely to have been designed to maximize the rents to Europeancolonists, not to maximize long-run growth.
The organization of society and institutions also persist (see,for example, the evidence presented in Acemoglu, Johnson, andRobinson [2001a]). Therefore, the institutions hypothesis alsosuggests that societies that are prosperous today should tend tobe prosperous in the future. However, if a major shock disruptsthe organization of a society, this will affect its economic perfor-mance. We argue that European colonialism not only disruptedexisting social organizations, but led to the establishment of, orcontinuation of already existing, extractive institutions in previ-ously prosperous areas and to the development of institutions ofprivate property in previously poor areas. Therefore, Europeancolonialism led to an institutional reversal, in the sense that
1263REVERSAL OF FORTUNE
regions that were relatively prosperous before the arrival of Eu-ropeans were more likely to end up with extractive institutionsunder European rule than previously poor areas. The institutionshypothesis, combined with the institutional reversal, predicts areversal in relative incomes among these countries.
The historical evidence supports the notion that colonizationintroduced relatively better institutions in previously sparselysettled and less prosperous areas: while in a number of coloniessuch as the United States, Canada, Australia, New Zealand,Hong Kong, and Singapore, Europeans established institutions ofprivate property, in many others they set up or took over alreadyexisting extractive institutions in order to directly extract re-sources, to develop plantation and mining networks, or to collecttaxes.15 Notice that what is important for our story is not the“plunder” or the direct extraction of resources by the Europeanpowers, but the long-run consequences of the institutions thatthey set up to support extraction. The distinguishing feature ofthese institutions was a high concentration of political power inthe hands of a few who extracted resources from the rest of thepopulation. For example, the main objective of the Spanish andPortuguese colonization was to obtain silver, gold, and othervaluables from America, and throughout they monopolized mili-tary power to enable the extraction of these resources. The min-ing network set up for this reason was based on forced labor andthe oppression of the native population. Similarly, the BritishWest Indies in the seventeenth and eighteenth centuries werecontrolled by a small group of planters (e.g., Dunn [1972, Chs.2–6]). Political power was important to the planters in the WestIndies, and to other elites in the colonies specializing in planta-tion agriculture, because it enabled them to force large masses ofnatives or African slaves to work for low wages.16
What determines whether Europeans pursued an extractive
15. Examples of extraction by Europeans include the transfer of gold andsilver from Latin America in the seventeenth and eighteenth centuries and ofnatural resources from Africa in the nineteenth and twentieth centuries, theAtlantic slave trade, plantation agriculture in the Caribbean, Brazil, and FrenchIndochina, the rule of the British East India Company in India, and the rule of theDutch East India Company in Indonesia. See Frank [1978], Rodney [1972],Wallerstein [1974–1980], and Williams [1944].
16. In a different vein, Europeans running the Atlantic slave trade, despitetheir small numbers, also appear to have had a fundamental effect on the evolu-tion of institutions in Africa. The consensus view among historians is that theslave trade fundamentally altered the organization of society in Africa, leading tostate centralization and warfare as African polities competed to control the supplyof slaves to the Europeans. See, for example, Manning [1990, p. 147], and also
1264 QUARTERLY JOURNAL OF ECONOMICS
strategy or introduced institutions of private property? And whywas extraction more likely in relatively prosperous areas? Twofactors appear important.
1. The economic profitability of alternative policies. Whenextractive institutions were more profitable, Europeanswere more likely to opt for them. High population density,by providing a supply of labor that could be forced to workin agriculture or mining, made extractive institutionsmore profitable for the Europeans.17 For example, thepresence of abundant Amerindian labor in Meso-Americawas conducive to the establishment of forced labor sys-tems, while the relatively high population density in Af-rica created a profit opportunity for slave traders in sup-plying labor to American plantations.18 Other types ofextractive institutions were also more profitable indensely settled and prosperous areas where there wasmore to be extracted by European colonists. Furthermore,in these densely settled areas there was often an existingsystem of tax administration or tribute; the large popula-tion made it profitable for the Europeans to take control ofthese systems and to continue to levy high taxes (see, e.g.,
Wilks [1975] for Ghana, Law [1977] for Nigeria, Harms [1981]) for the Congo/Zaire, and Miller [1988] on Angola.
17. The Caribbean islands were relatively densely settled in 1500. Much ofthe population in these islands died soon after the arrival of the Europeansbecause of the diseases that the Europeans brought (e.g., Crosby [1986] andMcNeill [1976]). It is possible that the initial high populations in these islandsinduced the Europeans to take the “extractive institutions” path, and subse-quently, these institutions were developed further with the import of slaves fromAfrica. An alternative possibility is that the relevant period of institutionaldevelopment was after the major population decline, but the Caribbean still endedup with extractive institutions because the soil and the climate were suitable forsugar production, which encouraged Europeans to import slaves from Africa andset up labor-oppressive systems (e.g., Dunn [1972] and Engerman and Sokoloff[1997, 2000]).
18. The Spanish conquest around the La Plata River (current day Argentina)during the early sixteenth century provides a nice example of how populationdensity affected European colonization (see Lockhart and Schwartz [1983, pp.259–260] or Denoon [1983, pp. 23–24]). Early in 1536, a large Spanish expeditionarrived in the area, and founded the city of Buenos Aires at the mouth of the riverPlata. The area was sparsely inhabited by nonsedentary Indians. The Spaniardscould not enslave a sufficient number of Indians for food production. Starvationforced them to abandon Buenos Aires and retreat up the river to a post atAsuncion (current day Paraguay). This area was more densely settled by semi-sedentary Indians, who were enslaved by the Spaniards; the colony of Paraguay,with relatively extractive institutions, was founded. Argentina was finally colo-nized later, with a higher proportion of European settlers and little forced labor.
1265REVERSAL OF FORTUNE
Wiegersma [1988, p. 69], on French policies in Vietnam, orMarshall [1998, pp. 492–497], on British policies in India).
2. Whether Europeans could settle or not. Europeans weremore likely to develop institutions of private propertywhen they settled in large numbers, for the natural reasonthat they themselves were affected by these institutions(i.e., their objectives coincided with encouraging good eco-nomic performance).19 Moreover, when a large number ofEuropeans settled, the lower strata of the settlers de-manded rights and protection similar to, or even betterthan, those in the home country. This made the develop-ment of effective property rights for a broad cross sectionof the society more likely. European settlements, in turn,were affected by population density both directly and in-directly. Population density had a direct effect on settle-ments, since Europeans could easily settle in large num-bers in sparsely inhabited areas. The indirect effectworked through the disease environment, since malariaand yellow fever, to which Europeans lacked immunity,were endemic in many of the densely settled areas [Ace-moglu, Johnson, and Robinson 2001a].20
Table VII provides econometric evidence on the institutionalreversal. It shows the relationship between urbanization or popu-lation density in 1500 and subsequent institutions using threedifferent measures of institutions. The first two measures refer tocurrent institutions: protection against expropriation risk be-tween 1985 and 1995 from Political Risk Services, which approxi-mates how secure property rights are, and “constraints on theexecutive” in 1990 from Gurr’s Polity III data set, which can bethought of as a proxy for how concentrated political power isin the hands of ruling groups (see Appendix 2 for detailedsources). Columns (1)–(6) of Table VII show a negative relation-
19. Extraction and European settlement patterns were mutually self-rein-forcing. In areas where extractive policies were pursued, the authorities alsoactively discouraged settlements by Europeans, presumably because this wouldinterfere with the extraction of resources from the locals (e.g., Coatsworth [1982]).
20. European settlements shaped both the type of institutions that developedand the structure of production. For example, while in Potosı (Bolivia) miningemployed forced labor [Cole 1985] and in Brazil and the Caribbean sugar wasproduced by African slaves, in the United States and Australia mining companiesemployed free migrant labor and sugar was grown by smallholders in Queensland,Australia [Denoon 1983, Chs. 4 and 5]. Consequently, in Bolivia, Brazil, and theCaribbean, political institutions were designed to ensure the control of the labor-ers and slaves, while in the United States and Australia, the smallholders and themiddle class had greater political rights [Cole 1985; Hughes 1988, Ch. 10].
1266 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EV
IIU
RB
AN
IZA
TIO
N,
PO
PU
LA
TIO
ND
EN
SIT
Y,
AN
DIN
ST
ITU
TIO
NS
Dep
ende
nt
vari
able
is:
Ave
rage
prot
ecti
onag
ain
stex
prop
riat
ion
risk
,19
85–1
995
Con
stra
int
onex
ecu
tive
in19
90C
onst
rain
ton
exec
uti
vein
firs
tye
arof
inde
pen
den
ce
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
Pan
elA
:W
ith
out
add
itio
nal
con
trol
s
Urb
aniz
atio
nin
1500
�0.
107
�0.
001
�0.
154
�0.
037
�0.
132
0.01
8(0
.043
)(0
.059
)(0
.066
)(0
.098
)(0
.069
)(0
.103
)L
ogpo
pula
tion
den
sity
in15
00�
0.37
�0.
37�
0.49
�0.
40�
0.33
�0.
54(0
.10)
(0.1
5)(0
.15)
(0.2
5)(0
.15)
(0.2
8)R
20.
140.
160.
250.
120.
120.
180.
310.
160.
37N
um
ber
ofob
serv
atio
ns
4275
4241
8441
4285
42
Pan
elB
:C
ontr
olli
ng
for
lati
tud
e
Urb
aniz
atio
nin
1500
�0.
097
�0.
001
�0.
159
�0.
038
�0.
128
0.02
2(0
.042
)(0
.059
)(0
.067
)(0
.099
)(0
.070
)(0
.104
)L
ogpo
pula
tion
den
sity
in15
00�
0.31
�0.
34�
0.45
�0.
41�
0.30
�0.
54(0
.10)
(0.1
5)(0
.16)
(0.2
5)(0
.16)
(0.2
8)L
atit
ude
2.87
3.53
2.57
�1.
492.
63�
1.86
1.52
2.68
1.48
(1.4
8)(1
.25)
(1.4
1)(2
.38)
(2.0
1)(2
.34)
(2.5
4)(2
.17)
(2.4
6)R
20.
210.
240.
310.
130.
130.
190.
320.
170.
38N
um
ber
ofob
serv
atio
ns
4275
4241
8441
4284
42
Sta
nda
rder
rors
are
inpa
ren
thes
es.R
egre
ssio
ns
use
data
for
allf
orm
erco
lon
ies
for
wh
ich
info
rmat
ion
onu
rban
izat
ion
and
popu
lati
onde
nsi
tyin
1500
isav
aila
ble,
asex
plai
ned
inth
ete
xt.U
rban
izat
ion
in15
00is
perc
ent
ofth
epo
pula
tion
livi
ng
into
wn
sw
ith
5000
orm
ore
peop
le.P
opu
lati
onde
nsi
tyin
1500
isto
tal
popu
lati
ondi
vide
dby
arab
lela
nd
area
from
McE
vedy
and
Jon
es[1
978]
.A
vera
gepr
otec
tion
agai
nst
expr
opri
atio
nri
skis
anev
alu
atio
nof
the
risk
that
priv
ate
inve
stm
ents
wil
lbe
expr
opri
ated
byth
ego
vern
men
t.C
onst
rain
tson
the
exec
uti
veis
anas
sess
men
tof
the
con
stit
uti
onal
lim
itat
ion
son
exec
uti
vepo
wer
.Reg
ress
ion
sw
ith
con
stra
ints
onex
ecu
tive
infi
rst
year
ofin
depe
nde
nce
use
the
earl
iest
avai
labl
eda
teaf
ter
inde
pen
den
ce,
and
also
incl
ude
the
date
ofin
depe
nde
nce
asan
addi
tion
alre
gres
sor.
For
deta
iled
sou
rces
and
desc
ript
ion
sse
eA
ppen
dix
2.
1267REVERSAL OF FORTUNE
ship between our measures of prosperity in 1500 and currentinstitutions.21
It is also important to know whether there was an institu-tional reversal during the colonial times or shortly after indepen-dence. Since the Gurr data set does not contain information fornonindependent countries, we can only look at this after indepen-dence. Columns (7)–(9) show the relationship between prosperityin 1500 and a measure of early institutions, constraint on theexecutive in the first year of independence, from the same dataset, while also controlling for time since independence as anadditional covariate. Finally, the second panel of the table in-cludes (the absolute value of) latitude as an additional control,showing that the institutional reversal does not reflect somesimple geographic pattern of institutional change.
The institutions hypothesis, combined with the institutionalreversal, predicts that countries in areas that were relativelyprosperous and densely settled in 1500 ended up with relativelyworse institutions after the European intervention, and thereforeshould be relatively less prosperous today. The reversal in rela-tive incomes that we have documented so far is consistent withthis prediction.
Notice, however, that the institutions hypothesis and thereversal in relative incomes do not rule out an important role forgeography during some earlier periods, or working through insti-tutions. They simply suggest that institutional differences are themajor source of differences in income per capita today. First,differences in economic prosperity in 1500 may be reflecting geo-graphic factors (e.g., that the tropics were more productive thantemperate areas) as well as differences in social organizationcaused by nongeographic influences. Second and more important,as we emphasized in Acemoglu, Johnson, and Robinson [2001a],a major determinant of European settlements, and therefore ofinstitutional development, was the mortality rates faced by Eu-ropeans, which is a geographical variable. Similarly, as noted byEngerman and Sokoloff [1997, 2000], whether an area was suit-able for sugar production is likely to have been important in
21. When both urbanization and log population density in 1500 are included,it is the population density variable that is significant. This supports the inter-pretation that it was the differences between densely and sparsely settled areasthat was crucial in determining colonial institutions (though, again, this may alsoreflect the fact that the population density variable is measured with less mea-surement error).
1268 QUARTERLY JOURNAL OF ECONOMICS
shaping the type of institutions that Europeans introduced. How-ever, this type of interaction between geography and institutionsmeans that certain regions, say Central America, are poor todaynot as a result of their geography, but because of their institu-tions, and that there is not a necessary or universal link betweengeography and economic development.
V. INSTITUTIONS AND THE MAKING OF THE MODERN WORLD
INCOME DISTRIBUTION
V.A. Institutions and the Reversal
We next provide evidence suggesting that institutional dif-ferences statistically account for the reversal in relative incomes.If the institutional reversal is the reason why there was a rever-sal in income levels among the former colonies, then once weaccount for the role of institutions appropriately, the reversalshould disappear. That is, according to this view, the reversaldocumented in Figures I and II and Tables III, IV, V, and VIreflects the correlation between economic prosperity in 1500 and in-come today working through the intervening variable, institutions.
How do we establish that an intervening variable X is re-sponsible for the correlation between Z and Y? Suppose that thetrue relationship between Y, and X, and Z is
(1) Y � � � X � � Z � �,
where � and are coefficients and � is a disturbance term. In ourcase, we can think of Y as income per capita today, X as ameasure of institutions, and Z as population density (or urban-ization) in 1500. The variable Z is included in equation (1) eitherbecause it has a direct effect on Y or because it has an effectthrough some other variables not included in the analysis. Thehypothesis we are interested in is that � 0; that is, populationdensity or urbanization in 1500 affects income today only viainstitutions.
This hypothesis obviously requires that there is a statisticalrelationship between X and Z. So we postulate that X � � Z �v. To start with, suppose that � is independent of X and Z andthat v is independent of Z. Now imagine a regression of Y on Zonly (in our context, of income today on prosperity in 1500,similar to those we reported in Tables III, IV, V, and VI):
1269REVERSAL OF FORTUNE
Y � b � Z � u1. As is well-known, the probability limit of theOLS estimate from this regression, b, is
plimb � � � � .
So the results in the regressions of Tables IV, V, VI, and VII areconsistent with � 0 as long as � � 0 and � 0. In this case, wewould be capturing the effect of Z (population density or urban-ization) on income working solely through institutions. This is thehypothesis that we are interested in testing. Under the assump-tions regarding the independence of Z from � and �, and of X from�, there is a simple way of testing this hypothesis, which is to runan OLS regression of Y on Z and X:
(2) Y � a � X � b � Z � u2
to obtain the estimates a and b. The fact that � in (1) is indepen-dent of both X and Z rules out omitted variable bias, so plima �� and plimb � . Hence, a simple test of whether b � 0 is all thatis required to test our hypothesis that the effect of Z is through Xalone.
In practice, there are likely to be problems due to omittedvariables, endogeneity bias because Y has an effect on X, andattenuation bias because X is measured with error or correspondspoorly to the real concept that is relevant to development (whichis likely to be a broad range of institutions, whereas we only havean index for a particular type of institutions). So the above pro-cedure is not possible. However, the same logic applies as long aswe have a valid instrument M for X, such that X � � � M � , andM is independent of � in (1). We can then simply estimate (2)using 2SLS with the first-stage X � c � M � d � Z � u3. Testingour hypothesis that Z has an effect on Y only through its effect onX then amounts to testing that the 2SLS estimate of b, b, is equalto 0. Intuitively, the 2SLS procedure ensures a consistent esti-mate of �, enabling an appropriate test for whether Z has a directeffect.
The key to the success of this strategy is a good instrumentfor X. In our previous work [Acemoglu, Johnson, and Robinson2001a] we showed that mortality rates faced by settlers are a goodinstrument for settlements of Europeans in the colonies and thesubsequent institutional development of these countries. Thesemortality rates are calculated from the mortality of soldiers,bishops, and sailors stationed in the colonies between the seven-
1270 QUARTERLY JOURNAL OF ECONOMICS
teenth and nineteenth centuries, and are a plausible instrumentfor the institutional development of the colonies, since in areaswith high mortality Europeans did not settle and were morelikely to develop extractive institutions. The exclusion restrictionimplied by this instrumental-variables strategy is that, condi-tional on the other controls, the mortality rates of Europeansettlers more than 100 years ago have no effect on GDP per capitatoday, other than their effects through institutional development.This is plausible since these mortality rates were much higherthan the mortality rates faced by the native population who haddeveloped a high degree of immunity to the two main killers ofEuropeans, malaria and yellow fever.
Table VIII reports results from this type of 2SLS test usingthe log of settler mortality rates as an instrument for institu-tional development. We look at the same three institutions vari-ables used in Table VII: protection against expropriation riskbetween 1985 and 1995, and constraint on the executive in 1990and in the first year of independence. Panel A reports results fromregressions that enter urbanization and log population density in1500 as exogenous regressors in the first and the second stages,while Panel B reports the corresponding first stages. Differentcolumns correspond to different institutions variables, or to dif-ferent specifications. For comparison, Panel C reports the 2SLScoefficient on institutions with exactly the same sample as thecorresponding column, but without including urbanization orpopulation density.
The results are consistent with our hypothesis. In all col-umns we never reject the hypothesis that urbanization in 1500 orpopulation density in 1500 has no direct effect once we control forthe effect of institutions on income per capita, and the addition ofthese variables has little effect on the 2SLS estimate of the effectof institutions on income per capita. This supports our notion thatthe reversal in economic prosperity reflects the effect of earlyprosperity and population density working through the institu-tions and policies introduced by European colonists.
V.B. Institutions and Industrialization
Why did the reversal in relative incomes take place duringthe nineteenth century? To answer this question, imagine a soci-ety like the Caribbean colonies where a small elite controls all thepolitical power. The property rights of this elite are relatively wellprotected, but the rest of the population has no effective property
1271REVERSAL OF FORTUNE
rights. According to our definition, this would not be a societywith institutions of private property, since a broad cross section ofsociety does not have effective property rights. Nevertheless,when the major investment opportunities are in agriculture, thismay not matter too much, since the elite can invest in the land
TABLE VIIIGDP PER CAPITA AND INSTITUTIONS
Institutions asmeasured by:
Dependent variable is log GDP per capita (PPP) in 1995
Averageprotection against
expropriationrisk, 1985–1995
Constraint onexecutive in
1990
Constraint onexecutive in first
year ofindependence
(1) (2) (3) (4) (5) (6)
Panel A: Second-stage regressions
Institutions 0.52 0.88 0.84 0.50 0.37 0.46(0.10) (0.21) (0.47) (0.11) (0.12) (0.16)
Urbanization in 1500 �0.024 0.030 �0.023(0.021) (0.078) (0.034)
Log population densityin 1500
�0.08 �0.10 �0.13(0.10) (0.10) (0.10)
Panel B: First-stage regressions
Log settler mortality �1.21 �0.47 �0.75 �0.88 �1.81 �0.78(0.23) (0.14) (0.44) (0.20) (0.40) (0.25)
Urbanization in 1500 �0.042 �0.088 �0.043(0.035) (0.066) (0.061)
Log population densityin 1500
�0.21 �0.35 �0.24(0.11) (0.15) (0.17)
R2 0.53 0.29 0.17 0.37 0.56 0.26Number of observations 38 64 37 67 38 67
Panel C: Coefficient on institutions without urbanization or population density in 1500
Institutions 0.56 0.96 0.77 0.54 0.39 0.52(0.09) (0.17) (0.33) (0.09) (0.11) (0.15)
Standard errors are in parentheses. Dependent variable is log GDP per capita (PPP) in 1995. Themeasure of institutions used in each regression is indicated at the head of each column. Urbanization in 1500is percent of the population living in towns with 5000 or more people. Population density is calculated as totalpopulation divided by arable land area. Constraint on the executive in 1990, 1900, and the first year ofindependence are all from the Polity III data set. Regressions with constraint on executive in first year ofindependence use the earliest available date after independence, and also include the date of independenceas an additional regressor.
Panel A reports the second-stage estimates from an IV regression with first-stage shown in Panel B.Panel C reports second-stage estimates from the IV regressions, which do not include urbanization orpopulation density and which instrument for institutions using log settler mortality. Log settler mortalityestimates are from Acemoglu, Johnson, and Robinson [2001a]. For detailed sources and descriptions seeAppendix 2.
1272 QUARTERLY JOURNAL OF ECONOMICS
and employ the rest of the population, and so will have relativelygood incentives to increase output.
Imagine now the arrival of a new technology, for example, theopportunity to industrialize. If the elite could undertake indus-trial investments without losing its political power, we may ex-pect them to take advantage of these opportunities. However, inpractice there are at least three major problems. First, those withthe entrepreneurial skills and ideas may not be members of theelite and may not undertake the necessary investments, becausethey do not have secure property rights and anticipate that theywill be held up by political elites once they undertake theseinvestments. Second, the elites may want to block investments innew industrial activities, because it may be these outside groups,not the elites themselves, who will benefit from these new activ-ities. Third, they may want to block these new activities, fearingpolitical turbulence and the threat to their political power thatnew technologies will bring (see Acemoglu and Robinson [2000,2001]).22
This reasoning suggests that whether a society has institu-tions of private property or extractive institutions may mattermuch more when new technologies require broad-based economicparticipation—in other words, extractive institutions may be-come much more inappropriate with the arrival of new technolo-gies. Early industrialization appears to require both investmentsfrom a large number of people who were not previously part of theruling elite and the emergence of new entrepreneurs (see Enger-man and Sokoloff [1997], Kahn and Sokoloff [1998], and Rothen-berg [1992] for evidence that many middle-class citizens, innova-tors, and smallholders contributed to the process of earlyindustrialization in the United States). Therefore, there are rea-sons to expect that institutional differences should matter moreduring the age of industry.
If this hypothesis is correct, we should expect societies withgood institutions to take better advantage of the opportunity toindustrialize starting in the late eighteenth century. We can testthis idea using data on institutions, industrialization, and GDPfrom the nineteenth and early twentieth centuries. Bairoch[1982] presents estimates of industrial output for a number ofcountries at a variety of dates, and Maddison [1995] has esti-
22. In addition, industrialization may have been delayed in some cases be-cause of a comparative advantage in agriculture.
1273REVERSAL OF FORTUNE
mates of GDP for a larger group of countries. We take Bairoch’sestimates of U. K. industrial output as a proxy for the opportunityto industrialize, since during this period the United Kingdom wasthe world industrial leader. We then run a panel data regressionof the following form:
(3) yit � �t � �i � � � Xit � � � Xit � UKINDt � �it,
where yit is the outcome variable of interest in country i at datet. We consider industrial output per capita and income per capitaas two different measures of economic success during the nine-teenth century. In addition, �t’s are a set of time effects, and �i’sdenote a set of country effects, UKINDt is industrial output in theUnited Kingdom at date t, and Xit denotes the measure of insti-tutions in country i at date t. Our institutions variable is againconstraint on the executive from the Gurr Polity III data set. Asnoted above, this variable is available from the date of indepen-dence for each country. Since colonial rule typically concentratedpolitical power in the hands of a small elite, for the purpose of theregressions in this table, we assign the lowest score to countriesstill under colonial rule. The coefficient of interest is �, whichreflects whether there is an interaction between good institutionsand the opportunity to industrialize. A positive and significant �is interpreted as evidence in favor of the view that countries withinstitutions of private property took better advantage of the op-portunity to industrialize. The parameter � measures the directeffect of institutions on industrialization, and is evaluated at themean value of UKINDt.
The top panel of Table IX reports regressions of equation (3)with industrial output per capita as the left-hand-side variable(see the note to the table for more details). Column (1) reports aregression using only pre-1950 data. The interaction term � isestimated to be 0.132, and is highly significant with a standarderror of 0.26. Note that Bairoch’s estimate of total U. K. indus-trialization, which is normalized to 100 in 1900, rose from 16 to115 between 1800 and 1913. In the meantime, the U. S. per capitaproduction grew from 9 to 126, whereas India’s per capita indus-trial production fell from 6 to 2. Since the average differencebetween the constraint on the executive in the United States andIndia over this period is approximately 6, the estimate impliesthat the U. S. industrial output per capita should have increasedby 78 points more than India’s, which is over half the actualdifference.
1274 QUARTERLY JOURNAL OF ECONOMICS
In column (2) we extend the data through 1980, again with noeffect on the coefficient, which stays at 0.132. In columns (3) and(4) we investigate whether independence impacts on industrial-ization, and whether our procedure of assigning the lowest scoreto countries still under colonial rule may be driving our results. Incolumn (3) we include a dummy for whether the country is inde-pendent, and also interact this dummy with U. K. industrializa-tion. These variables are insignificant, and the coefficient on theinteraction between U. K. industrialization and institutions, �, isunchanged (0.145 with standard error 0.035). In column (4) wedrop all observations from countries still under colonial rule, andthis again has no effect on the results (� is now estimated to be0.160 with standard error 0.048).
In columns (5) and (6) we use average institutions for eachcountry, X� i, rather than institutions at date t, so the equationbecomes
yit � �t � �i � � � X� i � UKINDt � �it.
This specification may give more sensible results if either varia-tions in institutions from year to year are endogenous with re-spect to changes in industrialization or income, or are subject tomeasurement error. � is now estimated to be larger, suggestingthat measurement error is a more important problem than theendogeneity of the changes in institutions.
An advantage of the specification in columns (5) and (6) isthat it allows us to instrument for the regressor of interest X� i �UKINDt, using the interaction between U. K. industrializationand our instrument for institutions, log settler mortality Mi (sothe instrument here is Mi � UKINDt). Once again, institutionsmight differ across countries because more productive or other-wise different countries have different institutions, and in thiscase, the interaction between industrialization and institutionscould be capturing the direct effects of these characteristics oneconomic performance. To the extent that log settler mortality isa good instrument for institutions, the interaction between logsettler mortality and U. K. industrialization will be a good instru-ment for the interaction between institutions and U. K. industri-alization. The instrumental-variables procedure will then dealwith the endogeneity of institutions, the omitted variables bias,and also the attenuation bias due to measurement error. The
1275REVERSAL OF FORTUNE
TA
BL
EIX
TH
EIN
TE
RA
CT
ION
OF
U.
K.
IND
US
TR
IAL
IZA
TIO
NA
ND
INS
TIT
UT
ION
S
For
mer
colo
nie
s,u
sin
gon
lypr
e-19
50da
ta(1
)
For
mer
colo
nie
s,u
sin
gda
tath
rou
gh19
80(a
llda
ta)
(2)
For
mer
colo
nie
s,u
sin
gon
lypr
e-19
50da
ta(3
)
For
mer
colo
nie
s,u
sin
gon
lyda
tapr
e-19
50an
dfo
rin
depe
nde
nt
cou
ntr
ies
(4)
For
mer
colo
nie
s,w
ith
aver
age
inst
itu
tion
sfo
rea
chco
un
try,
usi
ng
only
pre-
1950
data
(5)
For
mer
colo
nie
s,w
ith
aver
age
inst
itu
tion
sfo
rea
chco
un
try,
usi
ng
only
pre-
1590
data
(6)
For
mer
colo
nie
s,w
ith
aver
age
inst
itu
tion
sfo
rea
chco
un
try,
inst
rum
enti
ng
usi
ng
sett
ler
mor
tali
ty,
only
pre-
1950
data
(7)
For
mer
colo
nie
s,w
ith
aver
age
inst
itu
tion
sfo
rea
chco
un
try,
inst
rum
enti
ng
usi
ng
sett
ler
mor
tali
ty,
only
pre-
1950
data
(8)
For
mer
colo
nie
s,w
ith
aver
age
inst
itu
tion
sfo
rea
chco
un
try,
inst
rum
enti
ng
usi
ng
sett
ler
mor
tali
ty,
only
pre-
1950
data
(9)
For
mer
colo
nie
s,w
ith
aver
age
inst
itu
tion
sfo
rea
chco
un
try,
inst
rum
enti
ng
usi
ng
sett
ler
mor
tali
ty,
only
pre-
1950
data
(10)
Pan
elA
:D
epen
den
tva
riab
leis
ind
ust
rial
prod
uct
ion
per
capi
ta
U.
K.
indu
stri
aliz
atio
n�in
stit
uti
ons
0.13
20.
132
0.14
50.
160
0.20
20.
206
0.16
80.
169
0.15
60.
158
(0.0
26)
(0.0
27)
(0.0
35)
(0.0
48)
(0.0
19)
(0.0
22)
(0.0
30)
(0.0
32)
(0.0
65)
(0.0
65)
Inst
itu
tion
s8.
97�
3.36
10.5
17.
48(2
.30)
(4.4
6)(3
.50)
(9.5
1)In
depe
nde
nce
�14
.3�
6.4
1.1
2.0
(22.
9)(1
1.4)
(12.
6)(1
4.2)
U.
K.
indu
stri
aliz
atio
n�in
depe
nde
nce
�0.
12�
0.04
20.
046
0.06
(0.2
1)(0
.12)
(0.1
3)(0
.17)
U.
K.
indu
stri
aliz
atio
n�la
titu
de0.
130.
12(0
.50)
(0.4
8)R
20.
750.
740.
750.
840.
890.
890.
880.
880.
870.
87N
um
ber
ofob
serv
atio
ns
5975
5932
5959
5959
5959
1276 QUARTERLY JOURNAL OF ECONOMICS
Pan
elB
:D
epen
den
tva
riab
leis
log
GD
Ppe
rca
pita
Log
U.
K.
indu
stri
aliz
atio
n�in
stit
uti
ons
0.07
80.
060
0.07
30.
079
0.13
50.
130
0.15
90.
150
0.11
60.
111
(0.0
22)
(0.0
17)
(0.0
27)
(0.0
25)
(0.0
21)
(0.0
26)
(0.0
32)
(0.0
38)
(0.0
67)
(0.0
73)
Inst
itu
tion
s�
0.02
7�
0.08
4�
0.10
�0.
11(0
.025
)(0
.028
)(0
.04)
(0.0
4)In
depe
nde
nce
0.67
0.12
0.10
0.01
9(0
.27)
(0.1
3)(0
.13)
(0.1
6)L
ogU
.K
.in
dust
rial
izat
ion
�in
depe
nde
nce
0.03
5�
0.00
8�
0.04
20.
016
(0.1
2)(0
.093
)(0
.11)
(0.1
4)L
ogU
.K
.in
dust
rial
izat
ion
�la
titu
de0.
420.
42(0
.49)
(0.5
4)R
20.
950.
920.
950.
960.
960.
960.
960.
960.
960.
96N
um
ber
ofob
serv
atio
ns
7913
179
4679
7979
7979
79
Sta
nda
rder
rors
are
inpa
ren
thes
es.
All
colu
mn
sre
port
pan
elre
gres
sion
sw
ith
cou
ntr
yan
dpe
riod
dum
mie
sin
clu
ded.
Dep
ende
nt
vari
able
inP
anel
Ais
indu
stri
alou
tpu
tpe
rca
pita
1750
–198
0fr
omB
airo
ch[1
982]
.Dep
ende
nt
vari
able
inP
anel
Bis
log
GD
Ppe
rca
pita
1830
–198
0fr
omM
addi
son
[199
5].T
he
inst
itu
tion
sva
riab
leis
“Con
stra
int
onth
eex
ecu
tive
,”w
hic
his
anas
sess
men
tof
the
con
stit
uti
onal
lim
itat
ion
son
exec
uti
vepo
wer
.Th
ein
depe
nde
nt
vari
able
ofin
tere
stis
tota
lU.K
.in
dust
rial
outp
ut
inte
ract
edw
ith
con
stra
int
onth
eex
ecu
tive
inea
chco
un
try
from
the
Pol
ity
III
data
set.
Th
em
ain
effe
ctof
inst
itu
tion
sis
eval
uat
edat
the
mea
nva
lue
ofU
.K.i
ndu
stri
aliz
atio
n.P
olit
yII
Ipr
ovid
esin
form
atio
non
lyfo
rin
depe
nde
nt
cou
ntr
ies;
ifa
cou
ntr
yw
asa
colo
ny
ata
part
icu
lar
date
,we
assi
gnth
elo
wes
tva
lue
ofco
nst
rain
tson
the
exec
uti
ve,w
hic
his
1.A
vera
gein
stit
uti
ons
are
calc
ula
ted
over
the
valu
esin
Pol
ity
III
for
1750
,18
00,
1830
,18
60,
1880
,19
13,
and
1928
.W
eh
ave
anu
nba
lan
ced
pan
elw
ith
the
foll
owin
gob
serv
atio
ns.
For
indu
stri
alou
tpu
tw
eh
ave
data
onA
ust
rali
a,B
razi
l,C
anad
a,In
dia,
Mex
ico,
New
Zea
lan
d,S
outh
Afr
ica,
and
the
Un
ited
Sta
tes.
Inth
epa
nel
regr
essi
ons
for
GD
Ppe
rca
pita
befo
re19
50,w
eh
ave
data
onth
ese
cou
ntr
ies
(exc
ept
Sou
thA
fric
a)pl
us
Arg
enti
na,
Ban
glad
esh
,Bu
rma/
Mya
nm
ar,C
hil
e,C
olom
bia,
Egy
pt,G
han
a,In
dia,
Indo
nes
ia,P
akis
tan
,Per
u,a
nd
Ven
ezu
ela.
Inad
diti
on,f
orth
ere
gres
sion
usi
ng
GD
Ppe
rca
pita
data
thro
ugh
1980
,we
are
also
able
toin
clu
deE
thio
pia,
Ivor
yC
oast
,K
enya
,Mor
occo
,Nig
eria
,Sou
thA
fric
a,T
anza
nia
,an
dZ
aire
.We
hav
eda
tafo
rth
efo
llow
ing
date
s:17
50,1
800,
1830
,186
0,18
80,1
913,
1928
,195
3,an
d19
80,a
lth
ough
not
for
all
cou
ntr
ies
for
all
date
s.F
orde
tail
edso
urc
esan
dde
scri
ptio
ns
see
App
endi
x2.
1277REVERSAL OF FORTUNE
2SLS estimates reported in columns (7) and (8) are very similar tothe OLS estimates in columns (5) and (6), and are highlysignificant.23
In columns (9) and (10) we add the interaction betweenlatitude and industrialization. This is useful because, if thereason why the United States surged ahead relative to Indiaor South America during the nineteenth century is its geo-graphic advantage, our measures of institutions might be proxy-ing for this, incorrectly assigning the role of geography to in-stitutions. The results give no support to this view: the esti-mates of � are affected little and remain significant, while theinteraction between industrialization and latitude is insignifi-cant. Panel B of Table IX repeats these regressions using log GDPper capita as the left-hand-side variable (the interaction termis now as Mi � ln(UKINDt) since the left-hand-side variable is logof GDP per capita). The results are broadly similar to those inPanel A.
Overall, these results provide support for the view that in-stitutions played an important role in the process of economicgrowth and in the surge of industrialization among the formerlypoor colonies, and via this channel, account for a significantfraction of current income differences.
VI. CONCLUSION
Among the areas colonized by European powers during thepast 500 years, those that were relatively rich in 1500 are nowrelatively poor. Given the crude nature of the proxies for prosper-ity 500 years ago, some degree of caution is required, but thebroad patterns in the data seem uncontroversial. Civilizations inMeso-America, the Andes, India, and Southeast Asia were richerthan those located in North America, Australia, New Zealand, or
23. Despite our instrumental-variables strategy, the interaction betweeninstitutions and the opportunity to industrialize may capture the possible inter-action between industrialization and some country characteristics correlated withour instrument. For example, with an argument along the lines of Nelson andPhelps [1966] or Acemoglu and Zilibotti [2001], one might argue that industrialtechnologies were appropriate only for societies with sufficient human capital, andthat there were systematic cross-country differences in human capital correlatedwith institutional differences. This interpretation is consistent with our approach,since the correlation between institutions and human capital most likely reflectsthe fact that in societies with extractive institutions the masses typically did notor could not obtain education. In other words, low levels of human capital mayhave been a primary mechanism through which extractive institutions delayedindustrialization.
1278 QUARTERLY JOURNAL OF ECONOMICS
the southern cone of Latin America. The intervention of Europereversed this pattern. This is a first-order fact, both for under-standing economic and political development over the past500 years, and for evaluating various theories of long-rundevelopment.
This reversal in relative incomes is inconsistent with thesimple geography hypothesis which explains the bulk of the in-come differences across countries by the direct effect of geo-graphic differences, thus predicting a high degree of persistencein economic outcomes. We also show that the timing and natureof the reversal do not offer support to sophisticated geographyviews, which emphasize the time-varying effects of geography.Instead, the reversal in relative incomes over the past 500 yearsappears to reflect the effect of institutions (and the institutionalreversal caused by European colonialism) on income today.
Why did European colonialism lead to an institutional rever-sal? And how did this institutional reversal cause the reversal inrelative incomes and the subsequent divergence in income percapita across the various colonies? We argued that the institu-tional reversal resulted from the differential profitability of alter-native colonization strategies in different environments. In pros-perous and densely settled areas, Europeans introduced ormaintained already-existing extractive institutions to force thelocal population to work in mines and plantations, and took overexisting tax and tribute systems. In contrast, in previouslysparsely settled areas, Europeans settled in large numbers andcreated institutions of private property, providing secure prop-erty rights to a broad cross section of the society and encouragingcommerce and industry. This institutional reversal laid the seedsof the reversal in relative incomes. But most likely, the scale ofthe reversal and the subsequent divergence in incomes are due tothe emergence of the opportunity to industrialize during thenineteenth century. While societies with extractive institutionsor those with highly hierarchical structures could exploit avail-able agricultural technologies relatively effectively, the spread ofindustrial technology required the participation of a broad crosssection of the society—the smallholders, the middle class, and theentrepreneurs. The age of industry, therefore, created a consid-erable advantage for societies with institutions of private prop-erty. Consistent with this view, we documented that thesesocieties took much better advantage of the opportunity toindustrialize.
1279REVERSAL OF FORTUNE
APPENDIX 1: URBANIZATION ESTIMATES
This is a shortened version of the Appendix in Acemoglu,Johnson, and Robinson [2001b].
1. Urbanization in 1500
Our base estimates for 1500 consist of Bairoch’s [1988] as-sessment of urbanization augmented by the work of Eggimann[1999]. Merging these two series requires us to convert Eggi-mann’s estimates, based on a minimum population threshold of20,000, into Bairoch-equivalent urbanization estimates, based ona minimum population threshold of 5000.
To construct our base data, we run a regression of Bairochestimates on Eggimann estimates for all countries where theyoverlap in 1900 (the year for which we have the largest number ofBairoch estimates for non-European countries). There are thir-teen countries for which we have good overlapping data. Thisregression yields a constant of 6.6 and a coefficient of 0.67.
We use these results to convert from Eggimann to Bairoch-equivalent urbanization estimates in Colombia, Ecuador, Guate-mala (and other parts of Central America), Mexico, and Peru inthe Americas. We also use this method for all North Africancountries and for India (and the rest of the Indian subcontinent),Indonesia, Malaysia, Laos, Burma/Myanmar, and Vietnam inAsia. See Appendix 2 for the precise numbers we use.
There are a number of countries for which Bairoch deter-mines that there was no real urbanization or no pre-European“settled agriculture.” In these cases, a reasonable interpretationof Bairoch is that there was no urban population using his defi-nition. In our baseline data we therefore assume zero urbaniza-tion for the following countries: Argentina, Brazil, Canada, Chile,Guyana, Paraguay, Uruguay, the United States, and Australia.
For countries where Bairoch determines there was some lowlevel of urbanization, associated with fairly primitive agriculture,he assesses that the urbanization rate was 3 percent. We use thisestimate for Cuba, the Dominican Republic, Haiti, and Jamaicain the Americas. We also use this estimate for Hong Kong, thePhilippines, and Singapore in Asia and for New Zealand. In theAppendix of Acemoglu, Johnson, and Robinson [2001b], wepresent qualitative evidence documenting the low levels of urban-ization in countries with assigned values of 0 percent or 3 percenturbanization in our baseline data.
1280 QUARTERLY JOURNAL OF ECONOMICS
While the data on sub-Saharan Africa are worse than for anyother region, it is clear that urbanization before 1500 was at ahigher level than North America or Australia (see the Appendixof Acemoglu, Johnson, and Robinson [2001b] for detailed discus-sion and sources). Given the weakness and incompleteness ofdata for sub-Saharan Africa, we do not include any estimates inour baseline urbanization data set. We do, however, include all ofsub-Saharan Africa in our baseline population density data.
We have checked the robustness of our results using alter-native methods of converting Eggimann estimates into Bairoch-equivalent numbers. We have calculated conversion ratios at theregional level (e.g., for North Africa and the Andean regionseparately). We have also constructed an alternative series usinga conversion rate of 2, as suggested by Davis’ and Zipf ’s Laws (seeBairoch [1988], Chapter 9.)24 We have also used Bairoch’s overallassessment of urbanization for broad regions, e.g., Asia, withoutthe more detailed information from Eggimann (see the Appendixin Acemoglu, Johnson, and Robinson [2001b] for more detail). Wehave also used estimates just from Bairoch, just from Eggimann,and just from Chandler. See Table IV for relevant regressions.
Our baseline estimates and the most plausible alternativeseries are shown in Appendix 2. We have also calculated urbani-zation rates for all European countries and non-European coun-tries that were never colonized. We have also checked Bairoch’sestimates carefully for these countries against the work of Bai-roch, Batou, and Chevre [1988], Chandler and Fox [1974], deVries [1984], and Hohenberg and Lees [1985]. Our discussion ofurbanization in European and never colonized countries is notreported here to conserve space, but it is available from theauthors.
2. Urbanization from 1500 to 2000
Eggimann’s data only cover countries that are now part ofthe “Third World.” He therefore does not provide any informationon the timing of urbanization changes in settler colonies. Bairochdoes have some information on urbanization in the United States,Canada, and Australia, but only from 1800 [Bairoch 1988, Table13.4, p. 221]. For a more complete picture of urbanization from800 to 1850 across a wide range of countries, we therefore rely
24. We are using a conservative version of Davis’ law. See the Appendix inAcemoglu, Johnson, and Robinson [2001b] for a more detailed discussion.
1281REVERSAL OF FORTUNE
primarily on Chandler’s estimates. We should emphasize, how-ever, that wherever there is overlapping information, these esti-mates are broadly consistent with the findings of Eggimann andBairoch.25 As before, we convert urban population numbers intourbanization using population estimates from McEvedy andJones [1978].
Chandler’s data enable us to see changes in urbanizationover time across countries, but because his series ends in 1850 (or1861 for the Americas), we cannot follow the most importanttrends into the twentieth century. In addition, Chandler’s dataare reported at 50-year intervals from 1700 (100-year intervalsbefore that), which is only enough to show the broad pattern.
We therefore supplement the analysis with data from twoother sources. The UN [1969] provides detailed urbanization datafrom 1920, focusing on localities with 20,000 or more inhabitants(i.e., the same criterion as Chandler uses outside of Asia). How-ever, this still leaves a gap between 1850 and 1920.
We complete this composite series using data from Mitchell[1993, 1995]. His urbanization data start in 1750, provide infor-mation every ten years from 1790 for most countries, and run to1980. The only disadvantage of this series is the relatively latestarting date. The criterion for inclusion in Mitchell’s series isalso a little different—cities that had at least 200,000 inhabitantsaround 1970—but this seems to produce broadly consistent esti-mates for overlapping observations. We use these data both tocomplete the Chandler series for Mexico, India, and the UnitedStates (see Figure IVa) and to provide alternative estimates forthe timing of urbanization changes within the Americas.
The data shown in Figure IVa are from Chandler (through1850), Mitchell (for 1900), and the UN (for 1920 and 1930),converted to Bairoch-equivalent units using the conservativeZipf-Davis adjustment (i.e., multiplying the estimates by 2).
25. The only point of disagreement is whether there was any urbanization inthe area now occupied by the United States in 1500. Chandler lists one town(Nanih Waiya) but does not give its population. He also does not indicate anyurbanization either before or after this date. Bairoch argues there was no pre-European urbanization and the latest archaeological evidence suggests villagesrather than towns [Fagan 2000]. We therefore follow Bairoch in assigning a valueof zero. For supportive evidence see Waldman [1985, p. 30].
1282 QUARTERLY JOURNAL OF ECONOMICS
AP
PE
ND
IX2:
VA
RIA
BL
ED
EF
INIT
ION
SA
ND
SO
UR
CE
S
Var
iabl
eD
escr
ipti
onS
ourc
e
Log
GD
Ppe
rca
pita
(PP
P)
in19
95L
ogar
ith
mof
GD
Ppe
rca
pita
,on
Pu
rch
asin
gP
ower
Par
ity
Bas
is,
in19
95.
Wor
ldB
ank,
Wor
ldD
evel
opm
ent
Indi
cato
rs,
CD
-Rom
,19
99.
Dat
aon
Su
rin
ame
isfr
omth
e20
00ve
rsio
nof
this
sam
eso
urc
e.L
ogG
DP
per
capi
tain
1900
and
1950
Log
arit
hm
ofG
DP
per
capi
tain
1900
and
1950
.M
addi
son
[199
5]fo
r19
50;
Bai
roch
[197
8]fo
r19
00.
Indu
stri
alou
tpu
tpe
rca
pita
Inde
xof
indu
stri
aliz
atio
nw
ith
Bri
tain
in19
00eq
ual
to10
0.B
airo
ch[1
982]
.
Tot
alU
.K
.in
dust
rial
outp
ut
Inde
xeq
ual
to10
0in
1900
.B
airo
ch[1
982]
.L
ogpo
pula
tion
den
sity
in1
A.D
.,10
00,
and
1500
(als
olo
gpo
pula
tion
in15
00an
dlo
gar
able
lan
din
1500
)
Log
arit
hm
ofpo
pula
tion
den
sity
(tot
alpo
pula
tion
divi
ded
byto
tal
arab
lela
nd)
in1
A.D
.,10
00,
1500
.
McE
vedy
and
Jon
es[1
978]
.
Urb
aniz
atio
nin
1960
and
1995
Per
cen
tof
popu
lati
onli
vin
gin
urb
anar
eas
in19
60an
d19
95,
asde
fin
edby
the
UN
(typ
ical
ly20
,000
min
imu
min
hab
itan
ts).
Wor
ldB
ank,
Wor
ldD
evel
opm
ent
Indi
cato
rs,
CD
-Rom
,19
99.
For
mor
ede
tail
,se
ep.
159
ofth
eW
orld
Ban
k’s
Wor
ldD
evel
opm
ent
Indi
cato
rs19
99(h
ard
copy
).U
rban
izat
ion
in10
00,
1500
,an
d17
00P
erce
nt
ofpo
pula
tion
livi
ng
inu
rban
area
sw
ith
apo
pula
tion
ofat
leas
t50
00in
1000
,15
00,
and
1700
.
Bai
roch
and
supp
lem
enta
lso
urc
es,
asde
scri
bed
inA
ppen
dix
1.
Eu
rope
anse
ttle
men
tsin
1800
and
1900
Per
cen
tof
popu
lati
onth
atw
asE
uro
pean
orof
Eu
rope
ande
scen
tin
1800
and
1900
.R
ange
sfr
om0
to0.
99in
our
base
sam
ple.
McE
vedy
and
Jon
es[1
978]
and
oth
erso
urc
esli
sted
inA
ppen
dix
Tab
le5
ofA
cem
oglu
,Jo
hn
son
,an
dR
obin
son
[200
0].
Ave
rage
prot
ecti
onag
ain
stex
prop
riat
ion
risk
,19
85–1
995
Ris
kof
expr
opri
atio
nof
priv
ate
fore
ign
inve
stm
ent
bygo
vern
men
t,fr
om0
to10
,w
her
ea
hig
her
scor
em
ean
sle
ssri
sk.
We
calc
ula
ted
the
mea
nva
lue
for
the
scor
esin
all
year
sfr
om19
85to
1995
.
Dat
ase
tob
tain
eddi
rect
lyfr
omP
olit
ical
Ris
kS
ervi
ces,
Sep
tem
ber
1999
.T
hes
eda
taw
ere
prev
iou
sly
use
dby
Kn
ack
and
Kee
fer
[199
5]an
dw
ere
orga
niz
edin
elec
tron
icfo
rmby
the
IRIS
Cen
ter
(Un
iver
sity
ofM
aryl
and)
.T
he
orig
inal
com
pile
rsof
thes
eda
taar
eP
olit
ical
Ris
kS
ervi
ces.
1283REVERSAL OF FORTUNE
AP
PE
ND
IX2
(CO
NT
INU
ED
)
Var
iabl
eD
escr
ipti
onS
ourc
e
Con
stra
int
onex
ecu
tive
in19
70,
1990
,an
dfi
rst
year
ofin
depe
nde
nce
Ase
ven
-cat
egor
ysc
ale,
from
1to
7,w
ith
ah
igh
ersc
ore
indi
cati
ng
mor
eco
nst
rain
ts.
Sco
reof
1in
dica
tes
un
lim
ited
auth
orit
y;sc
ore
of3
indi
cate
ssl
igh
tto
mod
erat
eli
mit
atio
ns;
scor
eof
5in
dica
tes
subs
tan
tial
lim
itat
ion
s;sc
ore
of7
indi
cate
sex
ecu
tive
pari
tyor
subo
rdin
atio
n.
Sco
res
of2,
4,an
d6
indi
cate
inte
rmed
iate
valu
es.
Pol
ity
III
data
set,
dow
nlo
aded
from
Inte
r-U
niv
ersi
tyC
onso
rtiu
mfo
rP
olit
ical
and
Soc
ial
Res
earc
h.
Var
iabl
ede
scri
bed
inG
urr
[199
7].
Per
cen
tof
Eu
rope
ande
scen
tin
1975
reli
gion
vari
able
sP
erce
nt
ofpo
pula
tion
that
was
Eu
rope
anor
ofE
uro
pean
desc
ent
in19
75.
Ran
ges
from
0to
1in
our
base
sam
ple.
McE
vedy
and
Jon
es[1
978]
.
Per
cen
tage
ofth
epo
pula
tion
that
belo
nge
din
1980
(or
for
1990
–199
5fo
rco
un
trie
sfo
rmed
mor
ere
cen
tly)
toth
efo
llow
ing
reli
gion
s:R
oman
Cat
hol
ic,
Pro
test
ant,
Mu
slim
,an
d“o
ther
.”
La
Por
taet
al.
[199
9].
Col
onia
ldu
mm
ies
Du
mm
yva
riab
lein
dica
tin
gw
het
her
cou
ntr
yw
asa
Bri
tish
,F
ren
ch,
Ger
man
,S
pan
ish
,It
alia
n,
Bel
gian
,D
utc
h,
orP
ortu
gues
eco
lon
y.
La
Por
taet
al.
[199
9].
Tem
pera
ture
vari
able
sT
empe
ratu
reva
riab
les
are
aver
age
tem
pera
ture
,m
inim
um
mon
thly
hig
h,
max
imu
mm
onth
lyh
igh
,m
inim
um
mon
thly
low
,an
dm
axim
um
mon
thly
low
,al
lin
cen
tigr
ade.
Par
ker
[199
7].
1284 QUARTERLY JOURNAL OF ECONOMICS
Hu
mid
ity
vari
able
sH
um
idit
yva
riab
les
are
mor
nin
gm
inim
um
,m
orn
ing
max
imu
m,
afte
rnoo
nm
inim
um
,an
daf
tern
oon
max
imu
m,
all
inpe
rcen
t
Par
ker
[199
7].
Soi
lqu
alit
yM
easu
res
ofso
ilqu
alit
y/cl
imat
ear
est
eppe
(low
lati
tude
),de
sert
(low
lati
tude
),st
eppe
(mid
dle
lati
tude
),de
sert
(mid
dle
lati
tude
),dr
yst
eppe
was
tela
nd,
dese
rtdr
yw
inte
r,an
dh
igh
lan
d.
Par
ker
[199
7].
Nat
ura
lre
sou
rces
Mea
sure
sof
nat
ura
lre
sou
rces
are
perc
ent
ofw
orld
gold
rese
rves
toda
y,pe
rcen
tof
wor
ldir
onre
serv
esto
day,
perc
ent
ofw
orld
zin
cre
serv
esto
day,
perc
ent
ofw
orld
silv
erre
serv
esto
day,
and
oil
reso
urc
es(t
hou
san
dsof
barr
els
per
capi
tato
day)
.
Par
ker
[199
7].
Coa
lD
um
my
vari
able
equ
alto
1if
cou
ntr
yh
aspr
odu
ced
coal
sin
ce18
00.
Wor
ldR
esou
rces
Inst
itu
te[1
998]
and
Ete
mad
and
Tou
tain
[199
1].
Lan
dloc
ked
Du
mm
yva
riab
leeq
ual
to1
ifco
un
try
does
not
adjo
inth
ese
a.P
arke
r[1
997]
.
Isla
nd
Du
mm
yva
riab
leeq
ual
to1
ifco
un
try
isan
isla
nd.
DK
Pu
blis
hin
g[1
997]
.
Lat
itu
deA
bsol
ute
valu
eof
the
lati
tude
ofth
eco
un
try,
scal
edto
take
valu
esbe
twee
n0
and
1,w
her
e0
isth
eeq
uat
or.
La
Por
taet
al.
[199
9].
Log
mor
tali
tyL
ogof
esti
mat
edse
ttle
rm
orta
lity.
Set
tler
mor
talit
yis
calc
ulat
edfr
omth
em
orta
lity
rate
sof
Eur
opea
n-bo
rnso
ldie
rs,s
ailo
rs,a
ndbi
shop
sw
hen
stat
ione
din
colo
nies
.It
mea
sure
sth
eef
fect
sof
loca
ldi
seas
eson
peop
lew
itho
utin
heri
ted
orac
quir
edim
mun
itie
s.
Ace
mog
lu,
Joh
nso
n,
and
Rob
inso
n[2
001a
],ba
sed
onC
urt
in[1
989]
and
oth
erso
urc
es.
1285REVERSAL OF FORTUNE
AP
PE
ND
IX3
Bas
eu
rban
izat
ion
esti
mat
ein
1500
Sou
rce
ofba
seu
rban
izat
ion
esti
mat
ein
1500
Urb
aniz
atio
nes
tim
ate
in15
00u
sin
gon
lyin
form
atio
nfr
omB
airo
ch
Urb
aniz
atio
nes
tim
ate
in15
00u
sin
gon
lyin
form
atio
nfr
omE
ggim
ann
Urb
aniz
atio
nes
tim
ate
in15
00u
sin
gon
lyin
form
atio
nfr
omC
han
dler
Dav
is-Z
ipf
adju
stm
ent
appl
ied
toE
ggim
ann
seri
es
Pop
ula
tion
den
sity
in15
00
Pop
ula
tion
den
sity
in15
00
Pop
ula
tion
den
sity
in15
00
For
mer
colo
nie
sin
clu
ded
inou
rba
sesa
mpl
efo
ru
rban
izat
ion
For
mer
colo
nie
sin
clu
ded
inba
sesa
mpl
efo
rpo
pula
tion
den
sity
but
not
for
urb
aniz
atio
n
Arg
enti
na
0.0
Bai
roch
0.0
0.0
0.0
0.0
0.11
An
gola
1.50
Su
dan
14.0
3A
ust
rali
a0.
0B
airo
ch0.
00.
00.
00.
00.
03B
aham
as1.
46S
uri
nam
e0.
21B
angl
ades
h8.
5E
ggim
ann
con
vert
edto
Bai
roch
9.0
2.9
.5.
823
.70
Bar
bado
s1.
46T
anza
nia
1.98
Bel
ize
9.2
Egg
iman
n(3
.8%
)co
nve
rted
toB
airo
ch
7.0
18.0
19.6
7.6
1.54
Ben
in4.
23T
ogo
4.23
Bol
ivia
10.6
Egg
iman
n(E
cuad
oran
dB
oliv
ia)
con
vert
edto
Bai
roch
12.0
6.0
.12
.00.
83B
otsw
ana
0.14
Tri
nid
adan
dT
obag
o1.
46
Bra
zil
0.0
Bai
roch
0.0
0.1
.0.
20.
12B
urk
ina
Fas
o4.
23U
gan
da7.
51
Can
ada
0.0
Bai
roch
0.0
0.0
0.0
0.0
0.02
Bu
run
di25
.00
Zai
re1.
50C
hil
e0.
0B
airo
ch0.
00.
00.
00.
00.
80C
amer
oon
1.50
Zam
bia
0.79
Col
ombi
a7.
9E
ggim
ann
con
vert
edto
Bai
roch
7.0
2.0
2.0
4.0
0.96
Cap
eV
erde
0.50
Zim
babw
e0.
79
Cos
taR
ica
9.2
Egg
iman
n(3
.8%
)co
nve
rted
toB
airo
ch
7.0
18.0
.7.
61.
54C
entr
alA
fric
anR
epu
blic
1.50
Dom
inic
anR
epu
blic
3.0
Bai
roch
3.0
0.0
.0.
01.
46C
had
1.00
1286 QUARTERLY JOURNAL OF ECONOMICS
Alg
eria
14.0
Egg
iman
nco
nve
rted
toB
airo
ch.
11.0
11.0
22.0
7.00
Com
oros
4.48
Ecu
ador
10.6
Egg
iman
n(E
cuad
oran
dB
oliv
ia)
con
vert
edto
Bai
roch
12.0
6.0
5.0
12.0
2.17
Con
go1.
50
Egy
pt14
.6E
ggim
ann
con
vert
edto
Bai
roch
.11
.912
.423
.810
0.46
Cot
ed’
Ivoi
re4.
23
Gu
atem
ala
9.2
Egg
iman
n(3
.8%
)co
nve
rted
toB
airo
ch
7.0
18.0
19.6
7.6
1.54
Dom
inic
a1.
46
Gu
yan
a0.
0B
airo
ch0.
00.
0.
0.0
0.21
Eri
tria
2.00
Hon
gK
ong
3.0
Bai
roch
3.0
0.0
0.0
0.0
0.09
Eth
iopi
a6.
67H
ondu
ras
9.2
Egg
iman
n(3
.8%
)co
nve
rted
toB
airo
ch
7.0
18.0
19.6
7.6
1.54
Gab
on1.
50
Hai
ti3.
0B
airo
ch3.
00.
0.
0.0
1.32
Gam
bia
4.23
Indo
nes
ia7.
3E
ggim
ann
(In
don
esia
and
Mal
aysi
a)co
nve
rted
toB
airo
ch
9.0
1.0
0.5
2.0
4.28
Gh
ana
4.23
Indi
a8.
5E
ggim
ann
con
vert
edto
Bai
roch
9.0
2.9
1.8
5.8
23.7
0G
ren
ada
1.46
Jam
aica
3.0
Bai
roch
3.0
0.0
.0.
04.
62G
uin
ea4.
23L
aos
7.3
Egg
iman
n(L
aos
and
Vie
tnam
)co
nve
rted
toB
airo
ch
9.0
10.0
10.0
20.0
1.73
Ken
ya2.
64
Sri
Lan
ka8.
5E
ggim
ann
con
vert
edto
Bai
roch
9.0
2.9
.5.
815
.47
Les
oth
o0.
49
Mor
occo
17.8
Egg
iman
nco
nve
rted
toB
airo
ch.
16.7
21.3
33.3
9.08
Mad
agas
car
1.20
Mex
ico
14.8
Egg
iman
nco
nve
rted
toB
airo
ch7.
012
.36.
524
.62.
62M
alaw
i0.
79
1287REVERSAL OF FORTUNE
AP
PE
ND
IX3
(CO
NT
INU
ED
)
Bas
eu
rban
izat
ion
esti
mat
ein
1500
Sou
rce
ofba
seu
rban
izat
ion
esti
mat
ein
1500
Urb
aniz
atio
nes
tim
ate
in15
00u
sin
gon
lyin
form
atio
nfr
omB
airo
ch
Urb
aniz
atio
nes
tim
ate
in15
00u
sin
gon
lyin
form
atio
nfr
omE
ggim
ann
Urb
aniz
atio
nes
tim
ate
in15
00u
sin
gon
lyin
form
atio
nfr
omC
han
dler
Dav
is-Z
ipf
adju
stm
ent
appl
ied
toE
ggim
ann
seri
es
Pop
ula
tion
den
sity
in15
00
Pop
ula
tion
den
sity
in15
00
Pop
ula
tion
den
sity
in15
00
For
mer
colo
nie
sin
clu
ded
inou
rba
sesa
mpl
efo
ru
rban
izat
ion
For
mer
colo
nie
sin
clu
ded
inba
sesa
mpl
efo
rpo
pula
tion
den
sity
but
not
for
urb
aniz
atio
n
Mal
aysi
a7.
3E
ggim
ann
(In
don
esia
and
Mal
aysi
a)co
nve
rted
toB
airo
ch
9.0
1.0
0.5
2.0
1.22
Mal
i1.
00
Nic
arag
ua
9.2
Egg
iman
n(3
.8%
)co
nve
rted
toB
airo
ch
7.0
18.0
19.6
7.6
1.54
Mau
rita
nia
3.00
New Z
eala
nd
3.0
Bai
roch
3.0
0.0
.0.
00.
37M
ozam
biqu
e1.
28
Pak
ista
n8.
5E
ggim
ann
con
vert
edto
Bai
roch
9.0
2.9
.5.
823
.70
Nam
ibia
0.14
Pan
ama
9.2
Egg
iman
n(3
.8%
)co
nve
rted
toB
airo
ch
7.0
18.0
19.6
7.6
1.54
Nep
al13
.99
Per
u10
.5E
ggim
ann
con
vert
edto
Bai
roch
12.0
5.8
2.5
11.6
1.56
Nig
er1.
00
Ph
ilip
pin
es3.
0B
airo
ch3.
00.
0.
0.0
1.68
Nig
eria
4.23
Par
agu
ay0.
0B
airo
ch0.
00.
0.
0.0
0.50
Rw
anda
25.0
0S
inga
pore
3.0
Bai
roch
3.0
0.0
0.0
0.0
0.09
Sw
azil
and
0.49
1288 QUARTERLY JOURNAL OF ECONOMICS
El
Sal
vado
r9.
2E
ggim
ann
(3.8
%)
con
vert
edto
Bai
roch
7.0
18.0
19.6
7.6
1.54
Sen
egal
4.23
Tu
nis
ia12
.3E
ggim
ann
con
vert
edto
Bai
roch
.8.
111
.316
.311
.70
Sie
rra
Leo
ne
4.23
Uru
guay
0.0
Bai
roch
0.0
0.0
.0.
00.
11S
outh
Afr
ica
0.49
U.
S.
A.
0.0
Bai
roch
0.0
0.0
0.0
0.0
0.09
St.
Lu
cia
1.46
Ven
ezu
ela
0.0
Bai
roch
0.0
0.0
.0.
00.
44S
t.V
ince
nt
1.46
Vie
tnam
7.3
Egg
iman
n(L
aos
and
Vie
tnam
)co
nve
rted
toB
airo
ch
9.0
10.0
2.0
20.0
6.14
St.
Kit
tsan
dN
evis
1.46
Ou
rba
seu
rban
izat
ion
esti
mat
esar
eco
nst
ruct
edu
sin
gin
form
atio
nfr
omB
airo
chan
da
con
vers
ion
from
Egg
iman
n’s
esti
mat
esto
Bai
roch
-equ
ival
ent
esti
mat
es(a
sex
plai
ned
inth
ete
xtan
dA
ppen
dix
1).B
airo
ch-o
nly
esti
mat
esu
se9
perc
ent
for
all
Asi
anco
un
trie
s,7
perc
ent
for
Cen
tral
Am
eric
aan
dC
olom
bia,
12pe
rcen
tfo
rA
nde
anco
un
trie
s,3
perc
ent
for
cou
ntr
ies
wit
hm
inim
alu
rban
izat
ion
,an
d0
perc
ent
for
all
oth
erco
un
trie
sin
our
base
sam
ple.
Egg
iman
n-o
nly
esti
mat
esar
en
otad
just
edto
Bai
roch
-equ
ival
ent
un
its,
and
we
use
zero
for
cou
ntr
ies
inh
isda
tase
tw
ith
out
any
urb
anpo
pula
tion
in15
00.C
han
dler
-on
lyes
tim
ates
are
not
adju
sted
toB
airo
ch-e
quiv
alen
tu
nit
s,an
dw
eu
sea
valu
eof
zero
for
cou
ntr
ies
that
are
inh
isda
tase
tan
dfo
rw
hic
hh
edo
esn
otin
dica
tean
yu
rban
popu
lati
onin
1500
.T
he
Dav
is-Z
ipf
adju
stm
ent
dou
bles
Egg
iman
n’s
esti
mat
esbu
tu
ses
alo
wes
tim
ate
for
Cen
tral
Am
eric
a(d
etai
lsar
ein
the
App
endi
xof
Ace
mog
lu,J
ohn
son
,an
dR
obin
son
[200
1b].
Pop
ula
tion
den
sity
nu
mbe
rsar
eca
lcu
late
dfr
ompo
pula
tion
inM
cEve
dyan
dJo
nes
[197
8].W
edi
vide
esti
mat
edpo
pula
tion
in15
00by
lan
dar
eain
1995
(fro
mW
orld
Ban
k[1
999]
),ad
just
edfo
rar
able
lan
dar
eau
sin
gth
ees
tim
ates
inM
cEve
dyan
dJo
nes
[197
8].W
her
eM
cEve
dyan
dJo
nes
[197
8]on
lypr
ovid
ea
regi
onal
popu
lati
ones
tim
ate,
we
use
thei
rre
gion
alla
nd
area
esti
mat
ead
just
edfo
rar
able
lan
d.In
som
eca
ses
McE
vedy
and
Jon
es[1
978]
only
prov
ide
regi
onal
esti
mat
esof
popu
lati
onin
1500
.W
eth
eref
ore
use
regi
onal
aver
ages
ofpo
pula
tion
den
sity
for:
Wes
tA
fric
a(S
eneg
al,
Gam
bia,
Gu
inea
,Sie
rra
Leo
ne,
Ivor
yC
oast
,Gh
ana,
Bu
rkin
aF
aso,
Tog
o,B
enin
,an
dN
iger
ia);
Wes
t-C
entr
alA
fric
a(C
amer
oon
,Cen
tral
Afr
ican
Rep
ubl
ic,G
abon
,Con
go,Z
aire
,an
dA
ngo
la);
Rw
anda
and
Bu
run
di;S
outh
-Cen
tral
Afr
ica
(Zam
bia,
Zim
babw
e,an
dM
alaw
i);S
outh
Afr
ica,
Sw
azil
and,
and
Les
oth
o;N
amib
iaan
dB
otsw
ana;
the
Sah
elS
tate
s(M
auri
tan
ia,M
ali,
Nig
er,
and
Ch
ad—
base
don
qual
itat
ive
evid
ence
we
assu
me
asl
igh
tly
hig
her
popu
lati
onde
nsi
tyin
Mau
rita
nia
);E
ritr
eaan
dE
thio
pia
(bas
edon
qual
itat
ive
evid
ence
we
assu
me
ah
igh
erpo
pula
tion
den
sity
inE
thio
pia)
;Cen
tral
Am
eric
a(G
uat
emal
a,B
eliz
e,E
lS
alva
dor,
Hon
dura
s,N
icar
agu
a,C
osta
Ric
a,an
dP
anam
a);G
uya
na
and
Su
rin
ame
are
calc
ula
ted
from
the
aver
age
for
all
the
Gu
yan
as;a
nd
Pak
ista
n,I
ndi
a,an
dB
angl
ades
har
eca
lcu
late
dfr
omth
eav
erag
efo
rth
eIn
dian
subc
onti
nen
t.T
he
popu
lati
onde
nsi
tyin
Uru
guay
isas
sum
edto
beth
esa
me
asin
Arg
enti
na
in15
00.
Sin
gapo
rean
dH
ong
Kon
gar
eas
sum
edto
hav
eth
esa
me
popu
lati
onde
nsi
tyas
the
Un
ited
Sta
tes
in15
00.
Sm
alle
rC
arib
bean
isla
nds
are
assu
med
toh
ave
the
sam
epo
pula
tion
den
sity
asth
eD
omin
ican
Rep
ubl
icin
1500
.A
peri
od(.
)de
not
esm
issi
ng
data
.F
orfu
rth
erdi
scu
ssio
nof
sou
rces
,se
eA
ppen
dix
1.
1289REVERSAL OF FORTUNE
DEPARTMENT OF ECONOMICS, MASSACHUSETTS INSTITUTE OF TECHNOLOGY
SLOAN SCHOOL OF MANAGEMENT, MASSACHUSETTS INSTITUTE OF TECHNOLOGY
DEPARTMENTS OF POLITICAL SCIENCE AND ECONOMICS, UNIVERSITY OF CALIFORNIA,BERKELEY
REFERENCES
Abu-Lughod, Janet L., Before European Hegemony: The World System A.D. 1250–1350 (Oxford, UK and New York, NY: Oxford University Press, 1989).
Acemoglu, Daron, Simon Johnson, and James A. Robinson, “The Colonial Originsof Comparative Development: An Empirical Investigation,” NBER WorkingPaper No. 7771, 2000.
Acemoglu, Daron, Simon Johnson, and James A. Robinson, “The Colonial Originsof Comparative Development: An Empirical Investigation,” American Eco-nomic Review, XCI (2001a), 1369–1401.
Acemoglu, Daron, Simon Johnson, and James A. Robinson, “Reversal of Fortune:Geography and Institutions in the Making of the World Income Distribution,”NBER Working Paper No. 8460, 2001b.
Acemoglu, Daron, and James A. Robinson, “Political Losers as a Barrier toEconomic Development,” American Economic Review Papers and Proceed-ings, XC (2000), 126–130.
Acemoglu, Daron, and James A. Robinson, “Economic Backwardness in PoliticalPerspective,” unpublished, 2001.
Acemoglu, Daron, and Fabrizio Zilibotti, “Productivity Differences,” QuarterlyJournal of Economics, CXVI (2001), 563–606.
Ades, Alberto F., and Edward L. Glaeser, “Evidence on Growth, Increasing Re-turns, and the Extent of the Market,” Quarterly Journal of Economics, CXIV(1999), 1025–1046.
Bairoch, Paul, “Les grandes tendances des disparites economiques nationalesdepuis la revolution industrielle,” in P. Bairoch and M. Levy Laboyer, eds.,Regional and International Disparities in Economic Development Since theIndustrial Revolution, 7th International Economic History Conference (Ed-inburgh: 1978).
——, “International Industrialization Levels from 1750 to 1980,” Journal of Eu-ropean Economic History, XI (1982), 269–333.
——, Cities and Economic Development: From the Dawn of History to the Present(Chicago, IL: University of Chicago Press, 1988).
Bairoch, Paul, Jean Batou, and Pierre Chevre, La Population des villes Euro-peenes de 800 a 1850: Banque de Donnees et Analyse Sommaire des Resultats(Geneva: Centre d’histoire economique Internationale de l’Universite de Ge-neve, Libraire Droz, 1988).
Bloch, Marc, Land and Work in Medieval Europe (New York: Harper & Row,1966).
Boserup, Ester, The Conditions of Agricultural Growth: The Economics of Agrar-ian Change under Population Pressure (Chicago, IL: Aldine Publishing Com-pany, 1965).
Braudel, Fernand, The Structures of Everyday Life: Civilization and Capitalism,Fifteenth–Eighteenth Century (Berkeley and Los Angeles, CA: University ofCalifornia Press, 1992).
Chandler, Tertius, Four Thousand Years of Urban Growth: An Historical Census(Lewiston, NY: St. David’s University Press, 1987).
Chandler, Tertius, and Gerald Fox, Three Thousand Years of Urban Growth (NewYork, NY: Academic Press, 1974).
Chaudhuri, Kirti N., Asia Before Europe: Economy and Civilization of the IndianOcean from the Rise of Islam to 1750 (New York, NY: Cambridge UniversityPress, 1990).
Coatsworth, John H., “The Limits of Colonial Absolutism: Mexico in the Eigh-teenth Century,” in Karen Spalding ed., Essays in the Political, Economic andSocial History of Latin America (Newark, DE: University of Delaware Press,1982).
1290 QUARTERLY JOURNAL OF ECONOMICS
——, “Notes on the Comparative Economic History of Latin America and theUnited States,” in Walter L. Bernecker and Hans Werner Tobler, eds., De-velopment and Underdevelopment in America: Contrasts in Economic Growthin North and Latin America in Historical Perspective (New York, NY: Walterde Gruyter, 1993).
Cole, Jeffrey A., The Potosi Mita, 1573–1700: Compulsory Indian Labor in theAndes (Palo Alto, CA: Stanford University Press, 1985).
Collins, K. J., and D. F. Roberts, eds., Capacity for Work in the Tropics (Cam-bridge, UK: Cambridge University Press, 1988).
Crosby, Alfred, Ecological Imperialism: The Biological Expansion of Europe 900–1900 (New York, NY: Cambridge University Press, 1986).
Curtin, Philip D., Death by Migration (Cambridge, UK: Cambridge UniversityPress, 1989).
De Long, J. Bradford, and Andrei Shleifer, “Princes and Merchants: EuropeanCity Growth before the Industrial Revolution,” Journal of Law and Econom-ics, XXXVI (1993), 671–702.
de Vries, Jan, The Economy of Europe in an Age of Crisis, 1600–1750 (Cambridge,UK: Cambridge University Press, 1976).
——, European Urbanization, 1500–1800 (Cambridge, MA: Harvard UniversityPress, 1984).
Denoon, Donald, Settler Capitalism: The Dynamics of Dependent Development inthe Southern Hemisphere (Oxford, UK: Clarendon Press, 1983).
Diamond, Jared M., Guns, Germs and Steel: The Fate of Human Societies (NewYork, NY: W.W. Norton & Co., 1997).
DK Publishing, World Atlas (New York, NY: DK Publishing, 1997).Duby, Georges, The Early Growth of the European Economy; Warriors and Peas-
ants from the Seventh to the Twelfth Century (Ithaca, NY: Cornell UniversityPress, 1974).
Dunn, Richard S., Sugar and Slaves: The Rise of the Planter Class in the EnglishWest Indies 1624–1713 (Chapel Hill, NC: University of North Carolina Press,1972).
Eggimann, Gilbert, La Population des villes des Tiers-Mondes, 1500–1950 (Geneva:Centre d’histoire economique Internationale de l’Universite de Geneve, Li-braire Droz, 1999).
Eltis, David, “The Total Product of Barbados, 1664–1701,” Journal of EconomicHistory, LV (1995), 321–336.
Engerman, Stanley L., “Notes on the Patterns of Economic Growth in the BritishNorth America Colonies in the Seventeenth, Eighteenth and NineteenthCenturies” in Disparities in Economic Development since the Industrial Rev-olution, Paul Bairoch and Maurice Levy-Leboyer, eds. (New York, NY: St.Martin’s Press, 1981).
Engerman, Stanley L., and Kenneth L. Sokoloff, “Factor Endowments, Institu-tions, and Differential Paths of Growth among New World Economies,” inS. H. Haber, ed., How Latin America Fell Behind (Stanford, CA: StanfordUniversity Press, 1997).
Engerman, Stanley L., and Kenneth L. Sokoloff, “Institutions, Factor Endow-ments, and Paths of Development in the New World,” Journal of EconomicPerspectives, XIV (2000), 217–232.
Etemad, Bouda, and Jean-Claude Toutain, World Energy Production 1800–1985(Geneva: Libraire Droz, 1991).
Fagan, Brian M., Ancient North America: The Archaeology of a Continent, thirdedition (London, UK: Thames & Hudson, 2000).
Frank, Andre Gunder, Dependent Accumulation and Underdevelopment (London,UK: Macmillan Press, 1978).
Galenson, David W., “The Settlement and Growth of the Colonies: Population,Labor and Economic Development,” in Stanley L. Engerman and Robert E.Gallman, eds., The Cambridge Economic History of the United States, VolumeI (New York, NY: Cambridge University Press, 1996).
Gallman, Robert E., “Economic Growth and Structural Change in the Long Nine-teenth Century,” in Stanley L. Engerman and Robert E. Gallman, eds., TheCambridge Economic History of the United States, Volume II (New York, NY:Cambridge University Press, 2000).
1291REVERSAL OF FORTUNE
Gurr, Ted Robert, “Polity II: Political Structures and Regime Change, 1800–1986,” unpublished paper, Boulder, CO, University of Colorado, 1997.
Hall, Robert E., and Charles I. Jones, “Why Do Some Countries Produce so MuchMore Output per Worker than Others?” Quarterly Journal of Economics,CXIV (1999), 83–116.
Harms, Robert C., River of Wealth, River of Sorrow: The Central Zaire Basin in theEra of the Slave and Ivory Trade, 1500–1891 (New Haven, CT: Yale Univer-sity Press, 1981).
Hayek, Freidrich von, The Constitution of Liberty (Chicago, IL: University ofChicago Press, 1960).
Hodgson, Marshall G. S., Essays on Europe, Islam and World History (Cambridge,UK: Cambridge University Press, 1993).
Hohenberg, Paul M., and Lynn Hollen Lees, The Making of Urban Europe,1000–1950 (Cambridge, MA: Harvard University Press, 1985).
Hopkins, Anthony G., An Economic History of West Africa (New York, NY: Addi-son-Wesley Longman, 1973).
Hughes, Robert, The Fatal Shore (New York, NY: Vintage Books, 1988).International Labour Organization (ILO), Conditions of Work Digest, XIV (Geneva:
ILO, 1995).Jones, Charles I., “Population and Ideas: A Theory of Endogenous Growth,” NBER
Working Paper No. 6285, 1997.Kahn, Zorina, and Kenneth Sokoloff, “Patent Institutions, Industrial Organiza-
tion and Early Technological Change: Britain and the United States, 1790–1850,” in Maxine Berg and Kristine Bruland, eds., Technological Revolutionsin Europe: Historical Perspectives (Cheltenham, U.K. and Northampton, MA:Elgar, 1998).
Knack, Steven, and Philip Keefer, “Institutions and Economic Performance:Cross-Country Tests Using Alternative Measures,” Economics and Politics,VII (1995), 207–227.
Kremer, Michael, “Population Growth and Technological Change: One MillionB.C. to 1990,” Quarterly Journal of Economics, CVIII (1993), 681–716.
Kuznets, Simon, Modern Economic Growth: Rate Structure and Spread (NewHaven, CT: Yale University Press, 1968).
Landes, David S., The Wealth and Poverty of Nations: Why Some Are So Rich andSome So Poor (New York, NY: W.W. Norton & Co., 1998).
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W.Vishny, unpublished appendix from “The Quality of Government,” Journal ofLaw, Economics and Organization XV (1999), 222–279.
Law, Robin C. C., The Oyo Empire c1600–1836: A West African Imperialism in theEra of the Atlantic Slave Trade (New York, NY: Oxford University Press,1977).
Lewis, W. Arthur, Growth and Fluctuations 1870–1913 (London, UK: GeorgeAllen and Unwin, 1978).
Livi-Bacci, Massimo, A Concise History of World Population, third edition (Oxford,UK: Blackwell, 2001).
Locke, John, Two Treatises of Government (Indianapolis, IN: Hackett, 1690,1980).
Lockhart, James, and Stuart B. Schwartz, Early Latin America (New York, NY:Cambridge University Press, 1983).
Machiavelli, Niccolo, Discourses on Livy (New York, NY: Oxford University Press,1519, 1987).
Maddison, Angus, Monitoring the World Economy (Paris: OECD, 1820–1992,1995).
——, The World Economy: A Millennial Perspective (Paris: OECD, DevelopmentCentre of the Organization for Economic Cooperation and Development,2001).
Malthus, Thomas R., An Essay on the Principle of Population (Amherst, NY:Prometheus Books, 1798, 1998).
Manning, Patrick, Slavery and African Life: Occidental, Oriental and AfricanSlave Trades (New York, NY: Cambridge University Press, 1990).
Marshall, Alfred, Principles of Economics (London, UK: Macmillan, 1890).Marshall, P. J., “The British in Asia: Trade to Dominion, 1700–1765,” in P. J.
1292 QUARTERLY JOURNAL OF ECONOMICS
Marshall ed., The Oxford History of the British Empire, Volume II TheEighteenth Century (New York, NY: Oxford University Press, 1998).
McCusker, John J., and Russell R. Menard, The Economy of British America,1607–1785 (Chapel Hill, NC: University of North Carolina Press, 1985).
McEvedy, Colin, and Richard Jones, Atlas of World Population History (New York,NY: Facts on File, 1978).
McNeill, William H., Plagues and Peoples (Garden City, NJ: Anchor Press, 1976).——, A World History, fourth edition (Oxford, UK: Oxford University Press, 1999).Miller, Joseph C., Way of Death: Merchant Capitalism and the Angolan Slave
Trade, 1730–1830 (Madison, WI: University of Wisconsin Press, 1988).Mitchell, Brian R., International Historical Statistics, The Americas 1750–1988,
second edition (New York, NY: Stockton Press, 1993).Mitchell, Brian R., International Historical Statistics, Africa, Asia & Oceania
1750–1988, second edition (New York, NY: Stockton Press, 1995).Montesquieu, Charles de Secondat, The Spirit of the Laws (New York, NY:
Cambridge University Press, 1748, 1989).Myrdal, Gunnar, Asian Drama: An Inquiry into the Poverty of Nations, 3 volumes
(New York, NY: Twentieth Century Fund, 1968).Nelson, Richard, and Edmund Phelps, “Investment in Humans, Technological
Diffusion and Economic Growth,” American Economic Association Papers andProceedings, LVI (1966), 69–75.
North, Douglass C., Institutions, Institutional Change, and Economic Perfor-mance (New York, NY: Cambridge University Press, 1990).
North, Douglass C., and Robert P. Thomas, The Rise of the Western World: A NewEconomic History (Cambridge, UK: Cambridge University Press, 1973).
North, Douglass C., and Barry R. Weingast, “Constitutions and Commitment:Evolution of Institutions Governing Public Choice in Seventeenth CenturyEngland, Journal of Economic History, XLIX (1989), 803–832.
Notestein, Frank W., “Population: The Long View,” in Theodore W. Schultze ed.,Food for the World (Chicago, IL: University of Chicago Press, 1945).
Olson, Mancur, The Rise and Decline of Nations: Economic Growth, Stagflation,and Economic Rigidities (New Haven, CT: and London, UK: Yale UniversityPress, 1982).
——, Power and Prosperity: Outgrowing Communist and Capitalist Dictatorships(New York, NY: Basic Books, 2000).
Parker, Philip M., National Cultures of the World: A Statistical Reference, Cross-Cultural Statistical Encyclopedia of the World (Westport, CT: GreenwoodPress, 1997).
Pirenne, Henri, Medieval Cities: Their Origins and the Revival of Trade (NewYork, NY: Doubleday, 1956).
Pomeranz, Kenneth, The Great Divergence: Europe, China, and the Making of theModern World Economy (Princeton, NJ: Princeton University Press, 2000).
Postan, M. M., and E. E. Rich, The Cambridge Economic History of Europe:Volume 2, Trade and Industry in the Middle Ages (Cambridge, UK: Cam-bridge University Press, 1966).
Reid, Anthony, Southeast Asia in the Age of Commerce: Volumes 1 and 2, TheLands Below the Winds (New Haven, CT: Yale University Press, 1988 and1993).
Rodney, Walter, How Europe Underdeveloped Africa (London, UK: Bogle-L’Ouverture Publications, 1972).
Romer, Paul M., “Increasing Returns and Long-Run Growth,” Journal of PoliticalEconomy, XCIV (1986), 1002–1037.
Rothenberg, Winifred, The Transformation of Rural Massachusetts, 1750–1850(Chicago, IL: Chicago University Press, 1992).
Sachs, Jeffrey D., “Notes on a New Sociology of Economic Development,” inLawrence E. Harrison and Samuel P. Huntington, eds., Culture Matters: HowValues Shape Human Progress (New York, NY: Basic Books, 2000).
——, “Tropical Underdevelopment,” NBER Working Paper No. 8119, 2001.Showers, Victor, World Facts and Figures (New York, NY: Wiley, 1979).Simon, Julian L., The Economics of Population Growth (Princeton, NJ: Princeton
University Press, 1977).Smith, Adam, The Wealth of Nations (London, UK: Penguin Books, 1778, 1999).
1293REVERSAL OF FORTUNE
Tilly, Charles, Coercion, Capital, and European States, A.D. 990–1990 (Cam-bridge, MA: Basil Blackwell, 1990).
Tilly, Charles, and Wim P. Blockmans eds., Cities and the Rise of States in Europe,A.D. 1000 to 1800 (Boulder, CO: Westview Press, 1994).
Townsend, Richard F., The Aztecs (London, UK: Thames & Hudson, 2000).Toynbee, Arnold J., A Study of History, 12 Volumes (Oxford, UK: Oxford Univer-
sity Press, 1934–1961).United Nations (UN), Growth of the World’s Urban and Rural Population, 1920–
2000 (New York, NY: Department of Economic and Social Affairs, PopulationStudies, 1969).
——, World Urbanization Prospects: The 1996 Revision (New York, NY: Depart-ment of Economic and Social Affairs, Population Division, 1998).
Waldman, Carl, Atlas of the North American Indian (New York, NY: Facts on File,Inc., 1985).
Wallerstein, Immanuel M., The Modern World-System, 3 Volumes (New York, NY:Academic Press, 1974–1980).
Wiegersma, Nancy, Vietnam: Peasant Land, Peasant Revolution (New York, NY:St. Martin’s Press, 1988).
White, Lynn, Jr., Medieval Technology and Social Change (London, UK: OxfordUniversity Press, 1962).
Wilks, Ivor, Asante in the Nineteenth Century: The Structure and Evolution of aPolitical Order (New York, NY: Cambridge University Press, 1975).
Williams, Eric E., Capitalism and Slavery (Chapel Hill, NC: University of NorthCarolina Press, 1944).
World Bank, World Development Indicators (CD rom and book) (Washington, DC:World Bank, 1999).
World Coal Institute, Coal—Power for Progress, on the web at http://www.wci-coal.com/pfp.htm, March (2000).
World Resources Institute, World Resources: A Guide to the Global Environment,with The United Nations Environment Programme, The United NationsDevelopment Programme, and the World Bank (Oxford: Oxford UniversityPress, 1998).
Wrigley, Edward A., Continuity, Chance and Change (Cambridge, UK: CambridgeUniversity Press, 1988).
1294 QUARTERLY JOURNAL OF ECONOMICS
top related