9

The Diffusion of Institutions

Enrico Spolaore and Romain Wacziarg

Abstract
This chapter explores the fundamental drivers of economic development and political
institutions. It provides a novel empirical analysis of the determinants of institutional
differences and the diffusion of institutional innovations across societies. A critical dis-
cussion of the recent literature is presented, documenting how economic and political
outcomes are affected by traits that have deep historical and geographic roots and that
are passed on from generation to generation. The hypothesis is presented that intergen-
erationally transmitted traits affect current outcomes by acting as barriers to the diffu-
sion of technological and institutional innovations: a longer historical separation time
between populations creates greater barriers. Hence, the degree of ancestral distance
between a given society and the society at the frontier of institutional and technological
development should be associated with higher barriers and lower adoption. This hy-
pothesis is tested empirically with cross-country data. Empirical findings provide sub-
stantial support for the proposition that long-term historical distance from the frontier
affects both current institutions and development.

Introduction

Societies vary greatly not only in their levels of material prosperity but also in
the nature of their institutions, defined as the rules and norms which regulate
and constrain human actions and interactions. Why do we observe such vari-
ety, both over space and time? What are the fundamental drivers of economic
and institutional development across countries? Historians and social scientists
have been debating these important questions for centuries, emphasizing a wide
range of causes: from government policies and incentives to accumulate and
innovate, to geographic features and cultural factors. Building on the work of
geographers, anthropologists, and historians, economists have recently come
to recognize that factors deeply rooted in geography and history are important
determinants of economic and institutional development (e.g., Spolaore and
Wacziarg 2009; Putterman and Weil 2010; Comin et al. 2010; Ashraf and Galor
2013; for a synthesis of this literature, see Spolare and Wacziarg 2013).
146 E. Spolaore and R. Wacziarg

In this chapter, we provide new empirical evidence on the fundamental

drivers of development and institutions. We do not seek to explain the complex
interrelationships between institutions and development, the topic of a vast and
growing literature. Instead, we recognize that institutional development, char-
acterized by the protection of property rights, the enforcement of the rule of
law, and the adoption of democratic forms of government, is a key component
of modernity and development. We seek to characterize the deep roots of both
institutions and development, with a focus on the transmission of traits that are
passed on from one generation to the next over time.
At the center of our analysis is the distinction between direct effects and
barrier effects. In principle, traits passed on from generation to generation
throughout history could act directly to affect institutions and development. For
example, populations may directly inherit from their ancestors social norms
that facilitate cooperation and development. However, different populations
may also inherit traits that make them more or less similar to other populations,
in terms of habits, customs, language, and so on. Similarities or differences
in such traits could then facilitate or hinder exchanges between populations.
In particular, divergence in intergenerationally transmitted traits may act as
barriers to the adoption of novel institutions and technologies across different
societies. The main novelty of this analysis is that we seek to quantify the geo-
graphic and historical barriers to the adoption of institutions, defined primarily
as democratic control over political decision making. We show that these bar-
riers are both geographic and human/cultural, and highlight how the dynamics
of the diffusion of democracy are consistent with barrier effects.
We begin with a survey of the empirical evidence on the relationship be-
tween geography, deep history, institutions, and development. Thereafter we
discuss the mechanisms linking deep roots and current outcomes, developing
a taxonomy of possible effects, and conduct an empirical investigation of the
manner in which democratic institutions diffuse internationally.

The Relationship between Deeply Rooted

Factors, Institutions, and Development

Complexity and the Deep Roots of Institutions and Development

A brief account of recent views on the drivers of the wealth of nations must
stress the increasing complexity of the mechanisms highlighted in the econom-
ic literature. Decades ago, economists placed emphasis on proximate causes;
namely, the accumulation of factors of production (such as buildings and ma-
chines) and technological innovation. These factors, which were taken as ex-
ogenous, entered into an economys production function and thus represented
 The Diffusion of Institutions 147

the most proximate determinants of a societys income level.1 Later came a

recognition that these factors are endogenous, and that they depend on govern-
mental policies affecting the incentives to accumulate capital, to innovate, and
to adopt existing innovations.2 In an increasing focus on deeper determinants,
economists have come to recognize that policies themselves are endogenous;
they depend on political and social institutions, either formal or informal, that
create the right rules of the game, framing the decisions of policymakers and
economic actors (e.g., Acemoglu et al. 2001, 2002; Engerman and Sokoloff
1997; Greif 2006; North 1990, 2005; North and Weingast 1989; Ostrom 1990,
2005). The concept of institutions has become of paramount importance as a
driver of economic development in recent scholarship, although a debate rages
on regarding the direction of causality linking institutions and development.3
It is likely that causality runs both ways.
Over the last decade or so the economics literature has gone one step fur-
ther back and asked: What drives the deeper determinants of institutions? The
proposed answers center on geography and deep history, and even prehistory
(for review, see Spolaore and Wacziarg 2013). Figure 9.1 summarizes this ba-
sic schematic view of the causal links emphasized by economists. Researchers
have recognized the endogeneity of the proximate factors as well as policies and
institutions. Studies have increasingly emphasized deeper and more fundamen-
tal roots that are harder to change using policy or institutional reforms. These
deep roots include geographic factors and the legacy of past historical events.
The emphasis on deeper roots has created challenges of a different kind for
this research. The proliferation of candidate explanations for economic devel-
opment has challenged social scientists to articulate the interrelationships that
exist between these variables and to estimate these complex effects econo-
metrically. Figure 9.2 illustrates two major problems with the causal chain
identified above. First, causality could run in reverse, from development to
proximate factors, policies, and institutions, and from these factors among
themselves. This is the problem of endogeneity, typified by the solid backward
arrows in Figure 9.2. Second, more deeply rooted factors could affect develop-
ment directly, without necessarily going through a more proximate cause. For
instance the climate, a feature of geography, could affect development directly
by reducing productivity without necessarily working through institutions and
1
The Solow growth model epitomizes the theoretical foundation of this early work with exog-
enous accumulation and innovation (Solow 1956). The history of growth scholarship and its
effect on development policy is summarized in the early chapters of Easterly (2001).
2
For theoretical foundations of endogenous growth, see Romer (1990) and Lucas (1988). A
large literature around the same time began looking at the empirical determinants of growth,
starting with Baumol (1986) and Barro (1991).
3
Acemoglu et al. (2008) argue that democracy affects development and not the other way
around; in contrast Barro (1999) and Murtin and Wacziarg (2014) find evidence in favor of
Lipsets modernization hypothesis, i.e., that human capital and economic development lead to
more democracy.
148 E. Spolaore and R. Wacziarg

Accumulation
Policies of factors
Geography giving the
Institutions Technological Development
Deep history right
incentives innovation or
adoption

Figure 9.1 Drivers of development.

policies. This is the problem of exclusion restrictions, displayed as hatched

forward arrows in Figure 9.2.4
So far the literature has largely skirted these issues by estimating directly
the reduced form relationship between the deep roots of development (geog-
raphy and the legacy of historical events) and development itself. The cost of
this shortcut has been a relative lack of ability to clearly formulate mechanisms
whereby geographic and historical factors work to affect current development
outcomes. What have we learned from this research?
At this stage of our understanding of the wealth of nations, we must contend
with some salient facts (for further details, see Spolaore and Wacziarg 2009,
2012, 2013, 2014). First, there is a lot of persistence in development outcomes,
institutional features and technological sophistication. Past technological so-
phistication, going as far back as two thousand years, has predictive power for
current outcomes (Comin et al. 2010), and so does past institutional sophistica-
tion (Putterman and Weil 2010).5 Second, there has also been some dramatic
change: the emergence of modern modes of production, the spread of new
technologies, the diffusion of novel fertility behaviors and democratic institu-
tions. There have also been some reversals in the fortunes of countries, despite
the persistence of outcomes, and these reversals beg explanation. Third, both
persistence and change are associated with intergenerational links: the long-
term history of populations matters, suggesting that the factors which affected
past development have been passed down from generation to generation over
the long run. Fourth, there is compelling evidence that the manner in which
some populations, originating in Eurasia, maintained a persistent development
advantage, has to do with human and geographic barriers that prevented the
spread of innovations, institutions, and modern behaviors to other popula-
tions. Quantifying these barriers is a key first step toward developing policies
to overcome them. Our analysis here builds on these four insights to provide
a quantitative account of the drivers of development and institutions, with a
special focus on the role of long-term barriers to the diffusion of institutional
innovations. We continue with a discussion of recent empirical contributions
on geography and development, reversal of fortune, and the role of ancestry.
4
Econometric attempts to address these issues have traditionally relied on the use of instrumen-
tal variables; see Deaton (2010) for a critical review of this approach.
5
For an interesting archeological and historical analysis of the persistence of preindustrial agri-
cultural practices in the Mediterranean, from a long-term perspective, see Halstead (2014).
 The Diffusion of Institutions 149

Accumulation
Policies of factors
Geography giving the
Institutions Technological Development
Deep history right
incentives innovation or
adoption

Figure 9.2 Drivers of development (with complicating factors).

Geography

Our account of the deep roots of development begins with geography. It is

widely recognized that there is a strong relationship between geographic fac-
tors and current development outcomes. In Spolaore and Wacziarg (2013) we
showed that in a cross section of countries, four geographic variables account
for 44% of the variation in per capita income in the year 2005: absolute latitude
(by far the largest effect), the percentage of a countrys land area that is lo-
cated in tropical climates, a dummy for landlocked countries, and a dummy for
islands. One possible interpretation of this correlation is that geographic fea-
tures (climate, soil quality, etc.) have a direct effect on current productivity and
wealth (Sachs 2001). An alternative interpretation is that geographic factors act
indirectly, through their historical legacy. Such is the hypothesis in Diamonds
(1997) influential work. He posits that specific geographic features (the shape
and size of continents, the availability of domesticated animals, the variety of
plant species) affected the degree of economic and institutional sophistication
in the past. For instance, centralized states with organized political hierarchies
were overwhelmingly a feature of agrarian societies. These early advantages,
transmitted intergenerationally, created compounded and persistent advantages
still detectable today.
Systematic empirical evidence for the proposition that geography works
through its historical legacy, consistent with Diamonds hypotheses, is pro-
vided by Olsson and Hibbs (2005) and Ashraf and Galor (2011). Olsson and
Hibbs (2005) showed that a set of geographic variables, capturing Diamonds
early geographic advantages, have strong explanatory power in a regression
explaining contemporary outcomes. Ashraf and Galor (2011) showed a strong
correlation between population density in 1500 (a proxy for technological so-
phistication in Malthusian times) and the number of years since the agricultural
transition, itself a major legacy of early climatic and biogeographic character-
istics (climates conducive to sedentary agriculture, the diversity of plant and
animal species, and the shape and size of continents).
150 E. Spolaore and R. Wacziarg

In sum, the recent empirical literature, while confirming strong correlations

between geographical conditions and development, has tended to interpret them
not as direct causal effects of contemporary geography on current economic
outcomes, but as the result of indirect effects of geography on technological
and institutional outcomes, often going back centuries or even millennia, but
with persistent effects on productivity and development in more recent times.

Reversal of Fortune

Further empirical evidence in favor of the indirect interpretation of the ef-

fect of geography on current economic outcomes comes from the reversal of
fortune documented by Acemoglu et al. (2002). Table 9.1a, columns 1 and 2,
distils such evidence: population density in 1500, an indicator of technological
sophistication in Malthusian times, is strongly correlated negatively with per
capita income in 2005 in a cross section of countries excluding Europe; this ef-
fect is even stronger for former European colonies. Presumably, if geographic
features conferred a persistent direct advantage for productivity, it would have
done so both in the past and today, precluding this sort of reversal. Instead,
early geographic advantages could have translated into superior institutions,
and colonial powers may have set up different kinds of institutions in the loca-
tions they colonized: inclusive ones in locations conducive to settlement, and
extractive ones in inhospitable locations. This is precisely the interpretation of
Acemoglu et al. (2001, 2002). According to their view, the reversal of fortune
reflects the indirect effect of geography on institutions inherited from the colo-
nial era, and the especially strong magnitude of the reversal for former colonies
is consistent with this idea. In turn, the quality of institutions explains varia-
tion in current prosperity: the factors that caused high-quality institutions also
cause high per capita incomes.
However, along with inclusive institutions, colonial powers brought with
them other potential advantages for long-term development, including their
human capital, culture, values, norms, and attitudes (Glaeser et al. 2004). Thus,
it is not clear whether the reversal of fortune merely captures the indirect ef-
fect of geographic characteristics working through institutions alone, or the
broader legacy of other human traits. The rest of Table 9.1 provides evidence
consistent with this broader view. When looking at the correlation between
population density in 1500 and income per capita in 2005, the evidence for a
reversal disappears when considering non-European countries currently inhab-
ited by the descendants of the populations that lived there in 1500 (Table 9.1a,
column 5). The reversal is predominantly a feature of the subsample of coun-
tries not inhabited by their original dwellers (Table 9.1a, column 4). Moreover,
when adding the European countries to the sample (Table 9.1b), the evidence
of a reversal is actually reversed: we now uncover persistence, in the form of a
positive correlation between past and current prosperity (for further evidence,
see Chanda et al. 2014; for a review of this debate, see Diamond 2014).
Table 9.1 Reversal of fortune. Dependent variable: log per capita income 2005.
(1) Whole World (2) Former (3) Not Former (4) Nonindigenous (5) Indigenous
European Colony European Colony
(a) Without European countries
Log of population 0.246 0.393 0.030 0.232 0.117
density in year 1500 (3.304)*** (7.093)*** (0.184) (2.045)** (1.112)
Beta coefficient on 27.77% 47.88% 3.08% 32.81% 11.72%
1500 density
Observations 136 98 38 33 103
R-squared 0.077 0.229 0.001 0.108 0.014
(b) With European countries
Log of population 0.027 0.170 0.193
density in year 1500 (0.389) (2.045)** (2.385)**
Beta coefficient on 3.26% 22.34% 20.00%
1500 density
Observations 171 73 138
R-squared 0.001 0.050 0.040
Robust t statistics are given in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%.
All regressions include a constant term (estimates not reported).
151
152 E. Spolaore and R. Wacziarg

The evidence of persistence when including European countries in the

sample is suggestive of an important long-term role of European populations.
Empirical evidence on the effect of European colonization on current develop-
ment is provided by Easterly and Levine (2009). They construct a novel dataset
on shares of Europeans in different countries during the early stages of colo-
nization and find a strong and positive relationship between such shares and
current development. Intriguingly, the effect is even stronger when the share of
Europeans is relatively low. This fact is difficult to reconcile with the view that,
when Europeans settled in small numbers in a colony (say, because of high
mortality rates), their main legacy on the countrys development was negative,
via bad institutions (Acemoglu et al. 2001). Instead, this evidence is more
consistent with the view that Europeans, when they settled in a country, even
in small numbers, tended to bring traits that would positively affect economic
development and institutions in the longer run. Easterly and Levine find that
the effect of colonial European shares is robust to controlling for the current
share of Europeans, but becomes insignificant when controlling for current
levels of education. They interpret these results as suggestive of an important
role for human capital and human-capital-promoting institutions, consistent
with Glaeser et al. (2004) and Galor et al. (2009).
In sum, there is a general consensus about a secular reversal of fortune
across regions colonized by Europeans in modern times: regions which used
to have a higher population density and relatively higher technology around
1500 (e.g., Mexico and Peru) are relatively poorer than regions which used to
be less populated and economically advanced around 1500 (e.g., the area now
occupied by the United States and Canada). The reasons for such reversals are
controversial. Some economists (Acemoglu et al. 2002) emphasize the effects
of different institutions brought by the colonizers, whereas others emphasize
the effects of broader intergenerationally transmitted traits, such as human
capital and culture, which may include institutions but also other human and
societal traits.6

Role of Ancestry

The evidence on the reversal of fortune for newly settled countries and per-
sistence for countries inhabited by descendants of their aboriginal premodern
populations suggests that the characteristics of a population, rather than only
institutions, is of paramount importance to explain contemporary variation in
economic performance. Putterman and Weils (2010) empirical analysis elo-
quently demonstrates this point. For each country in the world they construct
an ancestry matrix providing the share of the population originating, since
6
This is part of a broader debate on the complex relation between institutions and cultural norms
and traits, which has received increasing attention in the economic literature, including by
leading economic historians such as North (2005) and Greif (2006).
 The Diffusion of Institutions 153

1500, from each of the other countries in the sample. They explore the role
of deep history summarized by (a) years since the agricultural transition and
(b) a variable capturing a countrys experience with centralized states (i.e.,
historical institutional sophistication). They show that deep history variables
have predictive power for contemporary per capita income. But their main
contribution is to show that this explanatory power increases greatly when
these variables are pre-multiplied by the ancestry matrix. That is, the variables
are corrected for each country to represent the weighted average of its cur-
rent populations ancestral historical variables. Thus, for instance, the United
States would appear with a short history of having a centralized state, but in
the ancestry adjusted version, the population of the United States comes from
countries which, in many cases, have had a much longer exposure to state
centralization. Thus, the history of populations, not of geographic locations,
explains contemporary economic performance.
In a closely related paper, Comin et al. (2010) document the strong autocor-
relation of technological sophistication going back one thousand years before
the Common Era. They develop several indices of technological advancement
based on the use of the most efficient and sophisticated technologies in 1000
BCE, 1 CE, 1500 CE, and today. In every past date considered, they show that
technological sophistication correlates significantly with current per capita in-
come or current technological advancement. This is strong evidence of persis-
tence. Going one step further, they show that this persistence is quantitatively
stronger when past measures of technological sophistication are adjusted for
ancestry in the same manner as Putterman and Weil (2010).
Finally, in two recent articles, Michalopoulos and Papaioannou (2013a,
b) provide evidence on the relations between precolonial ancestral traits and
economic development in Sub-Saharan Africa. They use the fact that national
boundaries in Africa are a recent outcome of colonization. Such artificial
borders partition over two hundred different ethnic groups across neighbor-
ing countries, thus subjecting them to different national institutions. On aver-
age, countries with better national institutions (in the sense of Acemoglu
et al. 2001) tend to have higher levels of development. However, the effect
of national institutions disappears when ethnically homogenous groups are
considered (Michalopoulos and Papaioannou 2013a): groups which share
common precolonial histories and cultures display a similar level of econom-
ic performance today, no matter whether they belong to African states with
better or worse institutions. This is strong evidence that intergenerationally
transmitted traits within African populations, going back to precolonial times,
are still key determinants of current development. When controlling for such
ancestral traits, national institutions, which reflect the legacy of more recent
colonial and postcolonial history, play only a marginal or even insignificant
role. Interestingly, Michalopoulos and Papaioannou (2013b) find that among
ancestral traits associated with current development, an important effect can be
attributed to measures of precolonial political centralization. Thus, institutions
154 E. Spolaore and R. Wacziarg

matter, but only those with deep roots at the ethnicity level (more akin to cul-
tural norms and traditions passed down from generation to generation) rather
than national-level institutions imposed more recently.
Overall, the evidence discussed here is consistent with the idea that traits
passed on from generation to generation within populationsincluding, but
not exclusively, a populations familiarity with certain types of political insti-
tutions inherited from deep historyare of paramount importance to explain
contemporary variation in economic development.

A Taxonomy of Mechanisms

Why do deep history and ancestry matter for economic and societal outcomes?
This question echoes a much broader and older debate about the effects of
intergenerationally transmitted traits. Historically, much of this discussion has
been on whether ancestry may matter because of nature, usually interpreted
in terms of biological inheritance, or nurture, including the transmission of
cultural traits (e.g., values and attitudes toward work, interpersonal trust or fer-
tility behavior). Most scholars of evolution today view the dichotomy between
nature and nurture as simplistic, because individuals inherit traits from their
ancestors through a complex interaction of different inheritance mechanisms,
biological and cultural, which interact with each other and with environmen-
tal and societal forces (for discussions, see Henrich and McElreath 2003;
Jablonka and Lamb 2005; Richerson and Boyd 2005). An instructive example
of geneculture interaction and coevolution is the spread of lactase persistence
(the genetically transmitted ability to digest milk in adulthood) in populations
which domesticated milk-producing animals (a culturally transmitted innova-
tion) (for in-depth analysis of the role of lactose tolerance in premodern devel-
opment, see Cook 2014). More generally, discussions of the relation between
evolution of human traits and economic and societal outcomes are provided,
for example, by Seabright (2010) and Bowles and Gintis (2011). An important
aspect of this evolutionary literature in economics is its focus on trust, reci-
procity, and cooperation, which are crucial mechanisms for the evolution of
institutions and their diffusion.
While the interactions among different inheritance mechanisms are of fun-
damental importance to understand the evolution of biological and cultural
traits, they are conceptually and empirically difficult to disentangle. Thus they
represent a major challenge when economists and other social scientists at-
tempt to identify links between specific intergenerational traits and specific
political and economic outcomes (see, e.g., Benjamin et al. 2012; Chabris et
al. 2013).
Rather than emphasizing the effects of different modes of transmission
(biological, cultural, or both), our discussion focuses on a different dimen-
sion of the relation between intergenerationally transmitted traits and societal
 The Diffusion of Institutions 155

outcomes: their mode of operation. How do inherited traits (whether bio-

logical, cultural, or both) affect economic and institutional development?
Somewhat surprisingly, there is not a lot of explicit analysis and discussion in
the economic literature about the different channels through which inherited
traits can, in principle, affect economic outcomes. This is in part because most
economists and other social scientists have implicitly focused only on direct
effects. That is, much of the research in this area has been about identifying
specific individual traits (e.g., work ethic or other intergenerationally transmit-
ted values) that directly affect income per capita and other economic outcomes.
Classic examples of these direct mechanisms would be the hypotheses about
the effects of the Protestant ethic on the emergence of capitalism formulated
by Weber (1905) or about the relationship between amoral familism and
underdevelopment in southern Italy (Banfield 1958).7
However, a direct effect of inherited traits is only one possible mechanism,
and not necessarily the most important one when considering technological
and institutional change in recent centuries. Another fundamental mechanism
is that similarity or divergence in inherited traits between populations can fa-
cilitate or hinder the diffusion of novel technologies, institutions, norms, and
behaviors across different societies. People that have inherited more similar
traits (e.g., norms, habits, language) may be able to communicate more easily
with each other, trust each other more, and learn more easily from each other,
while divergence in those traits may create barriers to the diffusion of novel
technologies and institutions across different societies.
Table 9.2 provides a summary of our taxonomy of the possible effects of
deeply rooted factors.
Barrier effects are at the center of our own research on the diffusion of
economic development and innovations (Spolaore and Wacziarg 2009, 2012,
2013, 2014).8 Our key hypothesis is that when populations enter into contact
with each other, differences in intergenerationally transmitted traits act as bar-
riers to exchange, communication, and imitation. In other words, the extent
to which different societies can learn from each other at a given time (i.e.,
the extent of horizontal transmission of cultural traits) should depend on the
7
Recent work by economists on the direct effect of culturally transmitted traits on economic
and institutional development includes Algan and Cahuc (2010) and Tabellini (2010). The
direct effects of genetic diversity within populations on comparative economic development
have been studied by Ashraf and Galor (2013). The relation between the emergence of post-
Malthusian growth and the evolution of specific traits (e.g., preferences for quantity vs. quality
of offspring) has been studied theoretically by Galor and Moav (2002) and historically by
Clark (2007). A selection and discussion of the literature on culture, ancestry, and economic
growth is provided in Spolaore (2014).
8
An important related literature focuses on the role of cultural barriers to economic exchanges.
For example, Guiso et al. (2009) find that more closely related populations in Europe tend to
trade more with each other because of higher trust among them. These findings are consistent
with our interpretation of the relationship between ancestry and barriers to economic and social
interactions across populations.
156 E. Spolaore and R. Wacziarg

Table 9.2 A taxonomy.

Type of Transmission Mechanism of Impact: Mechanism of Impact:

Direct Effect Barrier Effect
Biological: genetic or Galor and Moav (2002),
epigenetic Ashraf and Galor (2013)
Cultural: behavioral Weber (1905), Banfield (1958), Spolaore and Wacziarg
or symbolic Tabellini (2008) (2009, 2012, 2013, 2014)
Dual: geneculture Boyd and Richerson (1985),
interaction Cook (2014)

extent to which different societies have historically diverged in terms of verti-

cal transmission of traits across generations. We expect that these differences
are on average associated with measures of genetic distance between popula-
tions, because populations that are less closely related have had a longer time
to diverge along a larger range of ancestrally transmitted traits. However, the
relevant differences that constitute barriers to the spread of novel behavior and
innovations can reflect traits that are mostly transmitted culturally rather than
biologically, such as values, norms, preferences, and modes of communication.
Similarities and differences along these dimensions are likely to be very
relevant to explain the diffusion of complex technological and institutional
macro innovations, such as those at the roots of the Industrial Revolution,
which, according to Mokyr (2005:285), stemmed from novel ways to expand
the set of useful knowledge and applying natural philosophy to solve techno-
logical problems and bring about economic growth.
In our empirical work (Spolaore and Wacziarg 2009, 2012), we used mea-
sures of genetic distance from Cavalli-Sforza (1994) to test our model of diffu-
sion of economic development and technological innovations. In our empirical
work we use FST genetic distance, a commonly used metric of genetic distance
summarizing cross-population differences in genetic characteristics. This mea-
sure, based on a large number of neutral gene loci, is meant to measure separa-
tion times between populations (i.e., they measure the amount of time separat-
ing two current populations from a common ancestor population) and is used
to reconstruct phylogenies, the evolutionary history of human populations.9
We expect genetic distance to be associated with differences in a wide
range of human traits passed on both biologically and culturally from parents
to children, and therefore to capture the barriers or affinities that prevent these
populations from learning about each others institutions, technologies, and
behaviors. Genetic distance is meant to capture a wide range of mechanisms
that may account for barrier effects. For instance, genetic distance could pick
9
For cases of populations composed of different genetic groups, as is common in our new work
in particular, we calculate expected genetic distance between populations, by accounting for
the average distance between the groups composing each society.
 The Diffusion of Institutions 157

up distances in language, family networks, religious traditions, cultural prac-

tices, and other factors essential for establishing affinity and trust in reciprocal
human interactions. In our framework, these interactions promote diffusion,
and they are hindered by ancestral distance.
If barriers between societies play a central role in the diffusion of techno-
logical and institutional innovations, differences in income per capita or pro-
ductivity across societies should depend not so much on the absolute genetic
distance between them, but more on their relative genetic distance from the
innovators. For instance, to explain the spread of the Industrial Revolution in
the nineteenth century, what should matter is not so much the absolute distance
between two populationssay, the Italians and the Hungariansbut rather
how much closer the Italians were to the English than the Hungarians were.
In fact, we found that the magnitude of the effect of genetic distance relative
to the frontier (the English in the nineteenth century, the United States in the
twentieth century) is several times larger than that of absolute genetic distance.
In a race between the two measures (absolute and relative), the relative dis-
tance remained significant, but absolute genetic distance was not significantly
different from zero, consistent with our hypothesis. Moreover, as predicted
by a barriers model of diffusion, the effect of relative genetic distance from
the frontier rises during the nineteenth century along with the spread of the
Industrial Revolution. It reached a peak right before World War I and declined
thereafter as expected in a process of gradual diffusion, where, over time, in-
novations percolate and are adopted across increasingly more distant societies
(Figure 9.3).
The measure and estimation of barrier effects associated with ancestral dis-
tance in the spread of novel behavior are also at the center of our recent work
(Spolaore and Wacziarg 2014). In a sample of European populations and re-
gions from 1830 to 1970, we have empirically analyzed the determinants of
marital fertility using a novel data set of genetic and linguistic distances be-
tween European regions. Our key hypothesis is that the fertility decline can be
best understood as a process of diffusion of new social norms and behavioral
changes, spreading from early adopters to imitators through a mechanism of
social influence across time and space.10 In this respect, the fertility transition
was similar to the described spread of innovations associated with the diffu-
sion of the Industrial Revolution from England to other societies. However,
a key difference between the diffusion of fertility decline and the diffusion
of industrialization is that the two processes started at different frontiers. We
argue that the decline of fertility was associated with the diffusion of social and
behavioral changes from France, in contrast with the spread of the Industrial
10
Specifically, in our analysis of the diffusion of novel social norms associated with fertility de-
cline, we empirically test the implications of a theoretical model of social distance and social
influence, partly inspired by Akerlof (1997). More broadly, our approach is connected to a
large and growing interdisciplinary literature on social interactions and social networks (see,
e.g., Watts 2004; Jackson 2010; Ioannides 2012).
158 E. Spolaore and R. Wacziarg

19

relative genetic distance 17

Standardized beta on

15

13

11

7
1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year
Figure 9.3 Standardized effect of genetic distance on the absolute difference in per
capita income, over time, 18202005. Source: Authors calculations.

Revolution, where England played a leading role. Again, the mechanisms at

work can be interpreted in terms of barrier effects. The diffusion of the fer-
tility transition and the spread of industrialization followed different patterns
because societies at different relative distances from the respective innova-
torsthe French and the Englishfaced different barriers to social learning,
imitation, and adoption.
In sum, intergenerationally transmitted traits can potentially affect societal
outcomes through a diverse range of complex mechanisms, both in terms of
form of transmission (biological, cultural, dual) and mode of impact (directly
or as barriers). Recent empirical evidence on major societal changes, such as
the spread of industrialization and the diffusion of the fertility transition, sug-
gest a complex interaction between such channels. Divergence in ancestrally
transmitted traits may act as a temporary barrier to more recent social trans-
formation. Eventually, societies that are historically and culturally distant from
the innovation frontier also end up adopting novel norms and behaviors. Are
such barrier effects at work in the case of institutional diffusion and transfor-
mation? A first step toward addressing this question is taken in the next section.

The Diffusion of Institutions

The close connection between institutions and development, emphasized in the

economic literature (North and Weingast 1989; North 1990, 2005; Engerman
and Sokoloff 1997; Acemoglu et al. 2001) suggests that one of the channels by
which prosperity may spread is through the diffusion of novel institutions (e.g.,
democratic governance or the protection of property rights). There are several
reasons to believe that institutions may diffuse internationally. First, the transi-
tion toward democracy has occurred gradually, in hand with industrialization
 The Diffusion of Institutions 159

and the demographic transition (Murtin and Wacziarg 2014). As we have ar-
gued, industrialization and the demographic transition also diffused interna-
tionally from the innovation frontiers, in proportion to ancestral distance from
these frontiers (Spolaore and Wacziarg 2009, 2014). Second, democratization
tends to happen in waves and in regional clusters (Huntington 1993), suggest-
ing spillover effects from one location to another.
Is the spread of institutions from the institutional innovator (i.e., earlier
democracies such as England or the United States) also related to ancestral
distance? Is there a barrier effect induced by the degree of relatedness between
populations, limiting the spread of institutions to societies that are distant from
the institutional frontier?
To investigate these questions, we adopt a research design close to the
one we employed in our past research to study the spread of the Industrial
Revolution and of innovations.11 We seek to relate bilateral institutional dis-
tance to a country pairs relative distance to the institutional frontier. In the
following discussion, we use the United States as the frontier, but our results
are unchanged if we use England instead. To measure institutions, we use two
classes of variables, all measured for 1990 (to maximize the number of avail-
able observations): (a) a commonly used index of democracy (the Polity in-
dex), which captures constraints on the exercise of executive power and broad
political participation in competitive elections; (b) a series of subjective indi-
ces that capture economic institutions: the risk of repudiation of contracts, the
risk of expropriation, and an index of the rule of law.12 For each country pair,
we calculate the absolute average difference in each of these indices, a measure
of their institutional distance. We then regress these dependent variables on
relative genetic distance to the United States as well as a series of measures of
bilateral geographic barriers.13 The results are presented in Table 9.3.
Across all four measures of institutional quality, we find that relative genet-
ic distance bears a positive and statistically significant relationship with abso-
lute differences in institutions. We interpret this result as indicating that greater
separation times between populations introduce barriers to the adoption of bet-
ter institutions. In terms of magnitudes, the effect is sizable: a standard devia-
tion increase in genetic distance is associated with an increase in institutional
11
Our work is related to recent research on the links between culture and institutions (e.g., Ta-
bellini 2008, 2010). For a recent survey of the bidirectional causality between culture and
institutions, see Alesina and Giuliano (2015). However, these authors explored the direct link
between cultural traits and institutional features. In contrast, we seek to identify whether ances-
tral distance, which may partly or even predominantly capture distance in cultural traits, affects
the diffusion of institutions. In other words, we are concerned here with barrier effects.
12
Data source is the International Country Risk Guide, the same source sometimes used by Ac-
emoglu and his coauthors, and many other scholars, to capture institutional quality.
13
To account for spatial correlation in the dependent variable, we report standard errors that are
clustered two ways (at the level of the first and the second country in each pair), as has now
become common in this type of application.
160 E. Spolaore and R. Wacziarg

Table 9.3 Determinants of institutional differences.

(1) (2) (3) (4)

Democracy Repudiation Risk of Rule of Law
(Polity score) of Contracts Expropriation 1990
1990 1990 1990
Genetic distance relative 17.747** 7.846** 5.748* 3.693*
to the U.S.A., weighted (2.92) (3.36) (2.51) (2.53)
Absolute difference in 0.865 0.170 0.030 2.315**
latitudes (0.49) (0.32) (0.07) (5.63)
Absolute difference in 1.636 0.707 0.620 0.676*
longitudes (0.98) (1.33) (1.44) (2.09)
Geodesic distance 0.338 0.067 0.039 0.095*
(1000s of km) (1.83) (1.07) (0.77) (2.37)
1 for contiguity 2.790** 0.560** 0.493* 0.182
(4.94) (2.66) (2.26) (1.72)
=1 if either country is 0.041 0.092 0.052 0.151
an island (0.10) (0.60) (0.50) (1.67)
=1 if either country is 0.093 0.114 0.104 0.092
landlocked (0.31) (0.55) (0.45) (0.82)
=1 if pair shares at least 0.223 0.104 0.096 0.032
one sea or ocean (0.50) (0.74) (0.65) (0.39)
Constant 6.134** 1.995** 1.805** 1.134**
(9.32) (7.45) (6.93) (9.38)
# observations 8,911 6,216 6,216 6,216
(# countries) (134) (112) (112) (112)
R-squared 0.043 0.061 0.048 0.139
Standardized beta (%) 12.83 18.79 15.70 14.19
Dependent variables appear in column 2; t statistics are based on two-way clustered
standard errors in parentheses; * p < 0.05, ** p < 0.01).

differences contained between 12.8% and 18.8% of each of these dependent

variables standard deviation. Geographic barriers also matter. While results
vary somewhat across columns of Table 9.3 and across measures of distance,
we generally find that geographic proximity is associated with smaller differ-
ences in institutions.
To further investigate the existence of barrier effects, in Table 9.4 we ex-
ploit the time dimension. We now focus on the index of democracy (the Polity
index) since the other indicators are not available for as long a time span. We
seek to examine the diffusion of democracy around the third wave of democra-
tization (Huntington 1993). Table 9.4 illustrates the diffusion process. In 1960,
at the time of decolonization, many countries adopted nominally democratic
 The Diffusion of Institutions 161

institutions, only to revert soon thereafter to their institutional steady state

(Barro 1999). This was the case for many formerly colonized countries in Sub-
Saharan Africa. Correspondingly, ancestral distance to the institutional frontier
did not initially matter. By 1970, the fundamental determinants of democracy
(low levels of income and education, a history of authoritarianism) had reim-
posed themselves, and democracy levels fell throughout much of the formerly
colonized world. At this point, the effect of ancestral distance could be de-
tected. As the third wave began with the democratization of Portugal in 1974,
the standardized magnitude of the effect of genetic distance started to rise,
reaching a peak in 1990 when the third wave was itself in full swing. As in-
creasingly more countries, at successively greater ancestral distances from the
institutional frontier, joined the democratic bandwagon, the effect of genetic
distance fell again, although as of 2005 it has not yet disappeared. A prediction
of our diffusion model is that, insofar as ancestral distance captures barriers
across societies, ultimately it should no longer have an effect at the end of the
diffusion process.14
More work needs to be done to establish fully the proposition that insti-
tutions diffuse internationally in proportion to a societys distance from the
earlier institutional innovators (i.e., the institutional frontier). The empirical
results from Table 9.4 show, however, that at least when it comes to the most
recent wave of democratization, separation times from the institutional frontier
do predict the dynamics of democratization. As with other outcomes from our
past research, it is useful to think of the effect of deep-rooted factors on current
institutions as resulting from barrier effects between societies, resulting from
having undergone divergent histories.
Finally, as mentioned, there is an ongoing debate on the extent to which
specific institutions have had a causal effect on development (for further dis-
cussion, see Spolaore and Wacziarg 2013). Here we have not attempted to
estimate the effects of institutions on productivity or standards of living, but
have focused directly on the determinants of the spread of institutional features
(e.g., such as protection of property rights and political rights) associated with
modern economic and/or political development. We do not claim, however,
that these are the only institutions to have diffused across societies in recent
times, or that the adoption of Western-style institutions (e.g., representative de-
mocracy) is a necessary or sufficient condition for modern economic growth,
as shown by the experience of a few economically successful but undemo-
cratic countries such as China. In principle, our approach could indeed be used
to study the diffusion of alternative (and possibly economically deleterious)
14
This is precisely what we found for the diffusion of the fertility transition in Europe, where we
had the luxury of observing the entire span of the diffusion process, from the initial innovation
in France to the ultimate adoption by all regions of Europe by the second half of the twentieth
century (Spolaore and Wacziarg 2014).
Table 9.4 Determinants of institutional differences. Dependent variable: absolute difference in Polity 2 democracy score.
162

1960 1970 1980 1990 2000 2005

Genetic distance relative to 7.136 16.619* 25.478** 39.331** 16.563* 14.839*
the U.S.A., weighted (1.13) (2.29) (2.97) (4.49) (2.46) (2.36)
Absolute difference in latitudes 3.826* 7.399** 5.457** 3.132 4.074** 3.846**
(2.34) (3.71) (2.94) (1.65) (3.37) (3.13)
Absolute difference in longitudes 2.358* 3.839* 3.401* 3.195 1.625 1.027
(2.06) (2.55) (2.26) (1.76) (1.11) (0.74)
Geodesic distance 0.324* 0.510** 0.439* 0.458* 0.242 0.198
(1000s of km) (2.41) (2.79) (2.32) (2.26) (1.61) (1.38)
1 for contiguity 2.354** 1.562* 2.474** 3.821** 2.139** 2.348**
(3.58) (2.01) (3.69) (5.93) (3.80) (3.90)
= 1 if either country is an island 0.839 1.570* 1.078 0.631** 1.388** 1.397**
(1.48) (1.98) (1.62) (2.77) (5.48) (5.86)
= 1 if either country is landlocked 0.219 0.442 0.584 0.599 0.308 0.030
(0.51) (1.07) (1.55) (1.36) (0.41) (0.05)
= 1 if pair shares at least one sea or 0.566 0.174 0.011 0.225 1.177* 1.269**
ocean (1.47) (0.32) (0.03) (0.57) (2.49) (2.64)
Constant 8.047** 6.718** 6.835** 5.774** 6.825** 6.714**
(16.58) (10.65) (10.58) (8.43) (8.92) (9.01)
R-squared 0.018 0.053 0.048 0.089 0.038 0.037
Standardized beta (%) 4.327 9.796 14.45 24.10 11.24 9.954
t statistics based on two-way clustered standard errors in parentheses; * p < 0.05; ** p < 0.01; * p < 0.05. All regressions run on a sample of
4,560 country pairs from 96 countries.
 The Diffusion of Institutions 163

institutions, such as fascism or communism. We leave this intriguing line of

inquiry for further research.15

Conclusion

Much has been learned from the recent literature on the deep roots of insti-
tutions and development. There is substantial persistence. Societies that de-
veloped centralized states, in the wake of the adoption of agriculture in the
Neolithic era, transmitted these complex institutions to their descendants.
Societies that adopted sophisticated technologies early in history tended to
have the most advanced technologies in the preindustrial era, and then to de-
velop and adopt industrial modes of production sooner. Yet there has also been
a lot of change. Some fortunes were reversed in the wake of the colonial era,
although the reversal is a feature more of locations than of populations: loca-
tions now populated by the descendants of those with advanced early institu-
tions and technologies continue to enjoy the benefit of these early advantages.
Institutions and technologies have evolved to become vastly more complex,
and have diffused to locations and populations that did not benefit from initial
advantages.
How can we make sense of this persistence and change? A narrative that
relies exclusively on the persistent direct effects of early conditions on current
outcomes would lead to a pessimistic conclusion: since one cannot change
a societys history, what hope is there to alter the wealth of nations? Such a
narrative would also face difficulties in accounting for change. Instead, we
have focused on barrier effects. If ancestral differences in human traits intro-
duce barriers to the diffusion of institutions and technologies, there is hope
that these barriers can be overcome, and that, ultimately, even societies at a
high ancestral distance from the innovator can adopt modern institutions and
technologies.
The spread of economic development to East Asia is an example of how
societies historically and culturally distant from the European innovators man-
aged to adopt technological and institutional innovations in modern times. This
experience is indeed consistent with the existence of long-term historical and
cultural barriers, because such barriers operate on average, and societies may
develop traits that make them closer to the innovator, or may even be able
to sidestep barriers altogether thanks to historical contingencies. When Japan
developed and modernized, it became a cultural foothold for South Korea and
other neighbors. In contrast, North Korea illustrates how deleterious policies
15
The spread of nondemocratic and totalitarian ideologies and institutions has often been the
outcome of conflict and wars rather than of voluntarily adoption. Therefore, the study of their
diffusion requires an explicit focus on peaceful cultural and historical barriers as well as on
the determinants of conquests and wars. For an empirical analysis of the impact of cultural and
historical relatedness on international conflict, see Spolaore and Wacziarg (in press).
164 E. Spolaore and R. Wacziarg

and institutions can kill economic and political development in a society, irre-
spective of historical and cultural affinities. More broadly, the study of specific
instances of cultural and institutional change and diffusion across societies can
shed light on the mechanisms underlying the spread of political and economic
development.
In our past research, and in this chapter, we have documented how impor-
tant historical and cultural barriers can be quantitatively. Measures of ances-
tral distance to the innovation frontier matters for a wide range of outcomes:
the spread of the Industrial Revolution from England (Spolaore and Wacziarg
2009); the spread of the fertility transition from France (Spolaore and Wacziarg
2014); the more recent spread of modern technologies from advanced econ-
omies, chiefly the United States (Spolaore and Wacziarg 2012); and, docu-
mented here for the first time, the spread of democratic institutions during the
third wave of democratization. The dynamic pattern of the effect of ancestral
distance, in all these applications, is highly suggestive of a barrier effect: the
effect of genetic distance rises during the initial phase of the diffusion process
and falls later when a growing number of societies at successively greater ge-
netic distance from the innovator adopt the institution or technology.
Policies to foster the spread of institutions and technologies should focus
on reducing the barriers imposed by divergent historical paths across societies.
Globalization is likely the strongest force to reduce barriers, as movements
of goods, ideas, capital, and people help societies overcome the barriers to
modern institutions and modern technologies. However, research also sug-
gests that we cannot expect forces of globalization and economic integration
to work automatically, irrespective of local cultures, histories, and traditions.
The process of institutional and economic development is a long-term process
of complex adaptation, whereby diverse societies with different cultures and
histories gradually change in response to new internal and external incentives.
These incentives are affected not only by economic variables, but also by fac-
tors traditionally studied by historians, linguists, anthropologists, biologists,
political scientists, and sociologists. Future research needs to build on such
interdisciplinary approaches to shed further light on the specific mechanisms
that characterize economic and institutional development.