Geographical and Cultural Patterns in Cross-Border Mergers ...
Transcript of Geographical and Cultural Patterns in Cross-Border Mergers ...
Geographical and Cultural Patterns in Cross-Border
Mergers and Acquisitions: The Role of Experience
Massimo Del Gatto1
‘G.d’Annunzio’ University, LUISS Guido Carli, CRENoS
Carlo S. MastinuUniversity of Cagliari
Abstract
We exploit temporal and spatial correlation in a cross-border merger and acquisition (M&A) database withworldwide coverage to show that firms’ M&A choices are significantly influenced by experience effects related toknowledge accumulation. Our empirical strategy rests on the assumption that experience effects are expected tobe at play in M&A directed toward culturally contiguous countries (arguably driven by knowledge accumulation)and not in M&A deals in geographically contiguous countries (arguably driven by transport costs). The analysisis conducted on a bilateral measure of cultural distance (encompassing linguistic, religious, and genetic distanceindicators), which is meant to capture the idea of the cultural heritage originated by historic linkages.
Keywords: Cross-border M&A, Cultural Distance, Contiguity, Experience, Knowledge.JEL Codes: R12, F23, L22, D23
1Corresponding author. Contact details: Massimo Del Gatto; ‘G.d’Annunzio’ University, Department of Economics (DEc); VialePindaro, 42; 65127 - PESCARA (ITALY); [email protected]
1
1 Introduction
A number of studies have analysed the extent to which the reasons behind firms’ mergers and acquisitions (M&A)
choices (e.g. strategic competition and acquisition of markets, technologies and intangible assets in general, skills
and managerial practices, etc.) are captured by country-level measures of ‘distance’. While geographical and
cultural distance has received significant attention, the empirical evidence is not univocal. For example, Di Giovanni
(2005), Hijzen et al. (2008), and Hyun and Kim (2010) document negative effects for geographic distance (Hijzen
et al. (2008) also show that the relationship is less negative for horizontal mergers). Di Giovanni (2005) also
report positive effects associated with sharing a common language. Boschma et al. (2016) report positive effects
for geographical, industrial, organisational, and institutional proximity on domestic M&As in Italy. Coeurdacier et
al. (2009) find that geographical distance is a non-significant determinant of M&A flows in developed economies.
Blonigen et al. (2007) report strong evidence of significant and positive contiguity effects in US outbound foreign
direct investment (FDI). Di Guardo et al. (2015) reveal significant negative effects of geographical and cultural
distance on cross-border M&As. Ahern et al. (2015) find M&A flows are negatively correlated with cultural
distance.
By highlighting the shortcomings of such country-specific measures of distance (see Ambos and Hakanson, 2014;
Yildiz, 2014), a recent vein of literature provides a possible explanation for the inconclusiveness of results. The idea
is that the experience gained by the firm through its prior cross-border M&A deals can actually soften the effect
of cross-country diversity (Shenkar, 2012) on its future M&A choices. This places emphasis on distance between
entities, rather than between countries (Gaur, et al., 2007; Chapman, et al., 2008).
The relevance of the experience dimension as a determinant of firms’ international investment has been cited in
different theoretical contests (mainly at the intersection between international business, international economics,
and regional studies) since the beginning of the eighties, when the upsurge of cross-border M&As flows began.
A first vein of literature sees transaction costs as the main determinant of firms’ international activities. Due to
imperfections in knowledge markets, investing abroad entails search and information costs, as well as costs related
to bargaining and contract enforcement (Kogut and Zander, 1993). These costs grow with the extent of cultural
differences between the entities involved (e.g., Kogut and Singh, 1988). Multinational enterprises (MNEs) eliminate
transaction costs by absorbing them inside the hierarchy of the firm (Buckley and Casson, 1976). Together with
ownership and location, international business theory regards these internalisation advantages as one of the most
crucial perceived advantages of MNEs (Dunning, 1977).
A second line of research recognises that investing abroad always entails an evaluation of whether the firm is able
to establish and manage its operations (Guillen, 2003) based on its capabilities and knowledge of the host country
environment. On the one hand, knowledge enables firms to compete in both domestic and international markets (i.e.
knowledge-based theories). On the other hand, firms can acquire market knowledge through increasing involvement
in the foreign country, thereby improving their organisational capabilities. Such a process of organisational learning
can be related to the observation of the experiences of other organisations located in the same geographic region
(Baum et al., 2000) or to a process of experiential learning in which relevant knowledge is derived from past
experience (Levitt and March, 1988).
2
Hence, past M&A deals can foster a learning process (Cyert and March, 1963; Leung et al., 2005; Shenkar,
2012; Hutzschenreuter et al., 2011) leading to improvements in terms of technological routines and know-how,
intangible assets, marketing intensity, product portfolio management, and plant/equipment planning (Delios and
Beamish, 2001; Perkins, 2014). According to Chao and Kumar (2010), the obstacles faced by a firm entering a
new location can be understood in terms of its ‘marginal distance’ (encompassing all sorts of learning accumulated
through its prior international pursuits) more than in terms of the institutional or cultural distance between the
host and the home country. Other contributions (Barkema et al., 1996; Haleblian and Finkelstein, 1999) highlight
the positive effect of firm experience in the post M&A phase by, for example, increasing the firm’s capability to deal
with uncertainty in deal negotiations, resolve deadlocks, and reduce the risk of deal abandonment in subsequent
deals (Very et al., 1997). Popli et al. (2016) demonstrate how previous experience in culturally similar countries
generates a ‘cultural experience reserve’ (Luo and Shenkar, 2011), which mitigates the positive impact of cultural
differences on cross-border deal abandonment.
The Uppsala model (Carlson, 1974; Johanson and Vahlne, 1977) also provides a framework in which firm
internationalisation is a learning process wherein the firm enlarges its involvement in international activities at the
pace of its learning capacity. After investing in a given country, it acquires new information about that market
and its surroundings, and so its next investment is likely to be in that or a ‘similar’ country. Hence, firms would
choose their target countries according to their perceived distance from those markets (familiar markets are ‘close’,
whereas unknown markets are ‘distant’); such distance decreases over time as new information is acquired. There
is mixed empirical evidence regarding the effectiveness of this idea (Yildiz and Fey, 2016). Davidson (1980) shows
that US investment in Canada and the UK are well beyond what their market size, growth, tariffs, and geographical
proximity would predict, arguably due to cultural similarity, and firms prefer countries in which they are active to
those in which they are not. Benito and Gripsrud (1992) find no evidence of cultural and experience effects. Mitra
and Golder (2002) find that cultural distance is an insignificant factor, unlike knowledge. Chapman et al. (2008)
argue that cultural distance measures should take managers’ perceptions into account by considering previous
interactions with the host country.
In summary, while the occurrence of knowledge-related experience effects fostered by a firm’s foreign investment
history is recognised in several (but somewhat distant) contexts, empirical evidence is still scant. First, to the best
of our knowledge, the evidence is limited to investment originated in a single country, which prevents general
conclusions being drawn. In fact, we are still unable to verify whether having already invested in a given country
increases the probability of investing there again. Second, it is still unclear whether and how past investment
choices interact, through affecting firm knowledge, and the cultural sphere or experience is also likely to reduce the
attrition of (physical) geography. In both cases, the eventual attrition of distance on firms’ international investment
choices cannot be based on country-level measures only; firm experience must be taken into account. Third, if the
experience effects occur through the cultural sphere, the probability of engaging in foreign investment in a given
country should also depend on whether there has been investment in countries that are culturally similar to that
country.
To address these issues, we exploit cross-time and cross-space correlation in cross-border M&A data (drawn
from the Thomson Financial Security Database), with worldwide coverage, to investigate the eventual presence of
3
knowledge-related experience effects and uncover cross-patterns of (country-specific) cultural distance and (firm-
specific) experience.
Our empirical strategy rests on isolating two different types of M&A deals, depending on the type of costs they
are mostly driven by: spatial and knowledge costs. The first type concerns activities that are more intensive in
transportation (e.g. international importing and exporting along the supply chain). In these cases, the costs related
to geographical distance are particularly binding, so that firms tend to choose countries that are (geographically)
closest. Conversely, M&As related to activities that are particularly sensitive to knowledge costs will tend to be
oriented to culturally similar countries.
Disentangling these two types of deals helps us identify experience effects. In fact, while knowledge can be
accumulated, the burden of transport costs related to economic activity in a country can be compressed only by a
negligible fraction (e.g. changing storage and logistics service providers) through experience accumulated via past
investment. As a consequence, with experience effects at play, the occurrence of M&A deals in culturally contiguous
countries (arguably driven by knowledge costs) is expected to positively and significantly depend on the number of
past M&As operated in those countries and those that are culturally contiguous. On the other hand, the number
of past M&As operated in the same country or in its neighbour countries is not expected to influence the choice
to engage in an M&A deal in a geographically country contiguous (arguably driven by spatial costs). If this is the
case in the data, we can conclude that the experience effects are related to a process of knowledge accumulation.
Following Guiso et al. (2006) and Spolaore and Wacziarg (2016), we build a measure of bilateral cultural
distance (encompassing linguistic, religious, and genetic differences across populations) to capture the idea of the
cultural heritage originated by historic linkages.
The econometric analysis strongly supports the presence of experience effects associated with knowledge accu-
mulation. These effects are strong and persist after controlling for key country-level characteristics. In particular,
M&As in culturally contiguous countries occur despite higher tax and risk rates. Moreover, the probability of
engaging in M&A deals in a culturally (geographically) contiguous country is positively (negatively) related to
labour costs.
Overall, our analysis supports the idea that: i) firms’ M&A choices are significantly influenced by experience
effects related to knowledge accumulation; ii) to uncover such experience effects, we need to look where they are
potentially at play, that is, in knowledge-intensive deals/activities.
From a policy perspective, these results have strong implications for the current protectionism wave in the
FDI area. We show that experience plays an important role in bringing foreign capital to geographically distant
markets, usually at rather different stages of economic development, and that knowledge accumulation requires
time. Hence, the social costs of compressing the outward flows of foreign investment is higher in the case of
knowledge-intensive deals. As long as they are not selective, re- and back-shoring policies bring about a high risk
of knowledge dissipation.
The exposition proceeds as follows. Section 2 sets out the empirical strategy. Section 3 describes the data and
defines the variables used. The empirical analysis is reported in Sections 4 (benchmark results) and 5 (robustness
checks). Concluding remarks are presented in Section 6. Appendix A outlines the data generating process. Ap-
pendix B provides an in-depth description of the M&A flows in our dataset, also addressing their relationship with
4
geographical and cultural distance. Appendix C highlights the similarities and differences between our cultural
measures and comparable available measures.2
2 Empirical strategy
The literature discussed in the previous section recognises that a firm’s M&A history is likely to generate experience
effects insofar as it is related to knowledge accumulation along its various dimensions. In turn, the idea of knowledge
accumulation intersects the sphere of ‘cultural’ distance. By investing in a given country, the firm increases its
capacity to take advantage of further investment in that country or in those that are culturally ‘similar’. This
can modify, even substantially, the effect exerted by cultural distance on firms’ M&A choices. To operationalise
this idea, we follow an empirical strategy (Appendix A provides a description of a fully consistent data generating
process) grounded on disentangling between two types of M&A deals: those driven by spatial costs and those
driven by knowledge costs. The first type embraces activities that are intensive in transportation and, as such,
specifically affected by geographic distance. Indeed, while delivery costs have shrunk considerably over recent
decades, the frequency of shipments has also grown and the combination of frequency of transactions, storage,
inventory-handling, and other forms of logistics costs has determined an overall increase in the economic costs of
geographic distance in some industries (McCann, 1998; McCann and Shefer, 2004). In these cases, the costs related
to geographical distance are particularly binding, so that firms tend to choose countries that are (geographically)
as close as possible. The second type of deals refers to foreign investment in activities that are more sensitive to
knowledge costs. Consistent with the literature reviewed thus far, this type of M&A operations tends to be oriented
to culturally similar countries. Indeed, the literature has shown that the location of knowledge-intensive industries
is affected by the knowledge transmission mechanism. For example, Storper (1995, 1997) argues that innovative
industries and activities tend to concentrate when the innovation process spreads out through informal relations,
more than through traded and formalised interactions. Other contributions (e.g. Lundvall, 1992; Camagni, 1991;
Capello, 1999; Lawson, 2000) stress that knowledge transmission often involves interactive learning among actors
through cooperation or other joint activities, for example. Todtling et al (2006) highlight how high-tech firms tend
to combine knowledge sources from the region with those of national and international origin in their innovation
process. Even transferring ‘ready’ pieces of information or knowledge (i.e. licensing a specific technology, reading a
patent description, imitating other firms’ practices) is arguably affected by the extent of cultural distance between
actors. Disentangling these two types of deals helps identify experience effects under the two following hypotheses.
First, while knowledge can be accumulated, the burden of transport costs related to economic activity in a country
cannot be compressed, or only by a negligible fraction (e.g. by changing storage and logistics services provider),
through experience accumulated via past investment. Second, the cost-minimising location choice is a contiguous
country in both cases: culturally contiguous in one case, geographically contiguous in the other. Thus, as long
as firms’ realised deals in geographically (culturally) contiguous countries reveal spatial costs (knowledge costs)
minimising choices, M&As in geographically contiguous countries can be used as a sort of control group. On
2The full (195x195) matrix of bilateral distance measures (i.e., linguistic, religious and genetic distance, as well as the overall measure
of cultural distance) can be downloaded from http://docenti.unich.it/delgatto/delgatto_web/research.htm along with areplication package including original data and STATA files.
5
the one hand, with experience effects at play, the occurrence of M&A deals in culturally contiguous countries is
expected to positively and significantly depend on the number of past M&As operated in the those countries and/or
those that are culturally contiguous to those ones. On the other hand, the choice to engage in a M&A deal in a
geographically contiguous country is not expected to be influenced by the number of past M&As operated in that
country or in its neighbour countries. If this is the case in the data, we can argue that the experience effects are
related to a process of knowledge accumulation.
This suggests bringing to the data the following
Testing Hypothesis. The probability that firm h, located in country H, engages in an M&A in culturally con-
tiguous country F is positively affected by experience (i.e., firm h’s number of past M&As) in country F and in
those that are culturally contiguous to F . The probability that firm h, located in country H, engages in a new
M&A in the geographically contiguous country F is not affected by experience in country F and in those that are
geographically contiguous to F .
Insofar as the Testing Hypothesis proves true in the data, we can conclude, consistently with the literature
presented in Section 1, that the experience effects are related to a process of knowledge accumulation.
3 Data and variables
Our main data source is the Thomson Financial Security Database, which is extensively described in Brakman et
al. (2006). For all the M&As worldwide, the database reports country of origin, country of destination, year, date
of announcement, value of acquisition and International Standard Industrial Classification (ISIC) sector code for
both the acquiring and the target firm. We focus on cross-border M&As and exclude tax havens. Regarding the
reference period, we concentrate on 1985-2007 to avoid the potentially distortive effects of the 2008 economic crisis
which, as highlighted in Appendix B (Figure B.1), led to stagnation in both FDI and M&A international flows,
eventually associated to diversion effects related to cross-country asymmetric protectionism.3
A known problem with M&A data is the presence of M&A sequences realized by the same firm in the same
destination and announced on the same date. Since these observations usually correspond to the acquisition of
different branches of the same firm, we consider only the first observation in such cases (we drop 1054 observations),
with a value given by the sum of the values of the single deals in the sequence. In addition, we exclude the M&As
of firms investing in a single country (2244 observations).
The final dataset consists of 24402 deals realized by 17457 firms distributed across 21 (OECD) countries and
directed to 143 different countries. Around 73% of the deals occur between firms operating in the same Standard
Industrial Classification (SIC) two-digit sector (horizontal M&As). Table 1 reports the descriptive statistics for the
21 countries of origin. The US has the highest number of M&As, followed by the UK. However, it should be noted
that the UK is more involved in so-called ‘mega deals’, especially in the banking sector. This is also the case for
3The crisis hit differently the various economies, even turning into a double dip recession in some cases (e.g., in several Europeaneconomies). Although this heterogeneity might be controlled for in the estimates, by keeping the post-2008 years out of the analysis weaim at focusing on a phase characterized by a clear-cut positive trend in international M&A and FDI flows associated with increasingtrade freeness, thereby providing with quite general results.
6
France, which generates 4.52% of the total number of M&As but 10.23% of the total value. This phenomenon is
inverse in some countries, such as Ireland.
A more in-depth description of the dataset is reported in Appendix B.
Table 1 around here
Cultural distance (DcultH,F ) is a bilateral index obtained as a weighted average of three distance measures: linguistic,
religious, and genetic distance. The choice of these three dimensions is meant to capture the idea of the cultural
heritage originated by historic linkages and combines this idea with the genetic traits of populations. We follow
Guiso et al. (2006), who consider cultural distance as “those customary beliefs and values that ethnic, religious, and
social groups transmit fairly unchanged from generation to generation”, and Spolaore and Wacziarg (2016), who
posit that genetic distance is a summary statistic for a wide array of cultural traits transmitted inter-generationally.
Following Fearon’s (2003) approach, the three measures are constructed based on two building blocks: i)
the international distribution of languages/ethnic groups/religions (i.e. for each country, the percentage of the
population speaking each language/belonging to each ethnic group/professing each religion); ii) a distance matrix,
including all possible language/religion/ethnic group pairs, based on the dissimilarity between any two pairs in terms
of number of common branches in a ‘tree’. The term ‘branches’ describes the points where language/religion/ethnic
groups divide in the tree. The tree is a diagram showing the relationship between groups derived from a single
‘family’.
Following Fearon (2003), the index used to measure the distance between the two groups i and j is
τij = 1 −(l
m
)α(1)
where l is the number of shared branches between i and j; m is the maximum number of shared branches between
any two languages, religions, or ethnic groups; α is a parameter with an assigned value of 0.5.4
The distance between countries H and F is then calculated using the following formula:
DH,F =K∑k=1
(QHi Q
Fj τij
)k
(2)
where Qi and Qj denote the share of the population speaking languages i and j, belonging to ethnic groups i and
j, or professing religions i and j. K represents all possible combinations of languages, ethnic groups, or religions
in countries H and F . The index varies between 0 (minimal distance) and 1 (maximal distance).
We use Spolaore and Wacziarg’s (2009) approach to calculate genetic distance. Linguistic and religious distances
are our own calculation. For linguistic distance, the distance matrix is obtained by exploiting information in the
phylogenetic tree provided by ‘Ethnologue: Languages of the World’ ; also the international distribution of languages
is drawn from Ethnologue. For religious distance, we use the tree in Spolaore and Wacziarg (2009)5, while the
international distribution of religions is drawn from the CIA World Factbook.
4See Desmet et al. (2009, p.1301), for an explanation of the meaning and estimation of α.5We thank Roman Wacziarg and James Fearon for providing us with the tree
7
The final index of cultural distance is obtained as a weighted average of the three measures, with the weights
(0.27%, 0.39% and 0.34% for religious, linguistic and genetic distance, respectively) obtained through principal
component analysis.6
The correlation between geographic and cultural distance is quite low: 0.0523.
It is worth noting how several measures of cultural similarity are used in the literature. Di Guardo et al.
(2015) use the composite index proposed by Kaasa (2013) and Kaasa et al. (2013). The index is computed by
applying principal component analysis to items provided by the World and European Value Surveys relating to four
dimensions suggested by Hofstede (1980): power distance, uncertainty avoidance, individualism versus collectivism,
and masculinity versus femininity. Ahern et al. (2015) use information on trust, hierarchy and individualism drawn
from the World Values Survey. A widely used measure is the cultural distance index developed by Kogut and Singh
(1988). While such indexes differ in nature with respect to ours, comparable measures of cultural distance are those
provided by Melitz and Toubal (2014) and Spolaore and Wacziarg (2016): a comparison is provided in Appendix
C.
To conduct the analysis in Section 4, we need to set a criterion of cultural contiguity. Given the measure
of bilateral measure of cultural distance described in Section 3, we consider country H culturally contiguous to
country F if the cultural distance to F is, according to the bilateral measure described in Section 3, lower than the
value corresponding to the second percentile of country H’s distribution of bilateral distances and, in addition, lies
within the second decile of the worldwide distribution.7 (a different specification is used for robustness in Section
5).
Overall, M&As directed toward geographically and culturally contiguous countries represent about 17.5% and
32.5% of the sample, respectively, with only 5% of observations related to M&As in countries that are both
geographically and culturally contiguous. In the regression analysis, these observations are dropped to separate
out the geographical and cultural contiguity dimensions).
Information on geographical contiguity (i.e., sharing a common border) and geographic distance is taken from
the GeoDist database maintained by the Centre d’Etudes Prospectives et d’Informations Internationales – Paris
(CEPII). For geographic distance (i.e., DgeoH,F ), the variable dist is used. This is based on the simple geodesic
distance between the most populated cities/agglomerations in the two countries.8
CEPII data (namely, the Trade and Prod databases) are also used to obtain information on unit labour costs
(ULC ), calculated by applying wages to inverse labour productivity9. Average years of schooling (Schooling) are
from Barro and Lee (2013).
Among the other variables used, total population (Population), real interest rate (Intrate), and profit tax rate
6The complete replication package for the cultural index is available on http://docenti.unich.it/delgatto/delgatto_web/
research.htm.7Because of the combination of these two requirements, there might be countries whose culturally closest countries are not ‘close
enough’ to identify a culturally contiguous country, given the global distribution of bilateral distances. This is the case in Japan, forexample. Apart for these cases, the number of culturally contiguous countries identified ranges from one (Greece) to four (AU, CA,FN, SW, UK, and US). For the US, for instance, the culturally contiguous countries are, in order, AU, UK, NZ, and IR; For the UK,they are AU, NZ, IR, and the US.
8Geodesic distance is calculated following the great circle formula, which uses latitudes and longitudes of the most importantcities/agglomerations in terms of population.
9Using the CEPII variables notation, the ULC is calculated as wageva/lab
, where wage is the wage per employee, va is the value added,
and lab is the number of employees. The variables are expressed in nominal dollars.
8
(Tax ) are drawn from the World Development Indicators database provided by the World Bank.
Risk is the global index of country risk provided by SACE, which ranks all countries ranging from 1 (lowest
risk) to 9 (highest risk).10
4 Analysis
To gain insight into whether firms tend to reinvest in the same country and the extent to which this tendency
eventually changes in accordance with being the destination country geographically or culturally contiguous to the
country of origin, we report (Table 2) the results of a t-test statistics for the frequency of M&As directed towards:
countries in which the firm has already invested at least once (row 1), geographically (row 2) or culturally (row 3)
contiguous countries in which the firm has already invested at least once, second-order geographically (row 4) or
culturally (row 5) contiguous countries (i.e., being contiguous to contiguous country) in which the firm has already
invested at least once. The t-test considers the difference between the frequency observed in the original sample and
that observed in a control sample obtained by clustering the observations of the original sample into 553 groups,
including only M&As in given time windows by firms belonging to the same country and the same sector.11 The
results suggest the presence of a statistically significant tendency to reinvest in the same country. This tendency is
statistically significant in general (first row) and when restricted to M&As in contiguous countries (second, third,
and fourth rows). The null hypothesis of equal frequencies in the two samples is not rejected only in the case
of second-order cultural distance. However, this may depend on the very low share that second-order contiguous
countries represent, which shrink to 2.9% and 5%, respectively.
Table 2 around here
Such persistency in the data can be exploited to uncover possible patterns in firms’ M&A choices by bringing the
Testing Hypothesis stated in Section 2 to the data. Hence, we estimate the following equation:
mth(F conti H) = βo + β1P
th + β2M
th(F ) + β3M
th(Z contgeo F ) + β4M
th(Z contcult F ) + β5Γ
tH,F + εt
with i, j = geo, cult.
in which the dependent variable mth(F conti H) is a dichotomic variable assuming the value of 1 if the destination
country F is contiguous (geographically or culturally, depending on the specification) to the country of origin H,
and 0 otherwise. mth(F contcult H) = 1 in 3834 cases and mt
h(F contgeo H) = 1 in 1753 cases; that is, around 27
10The final ranking considers the political risk (associated with internal policy and international relationships), economic risk (eco-nomic conditions, public accounts, inflation, current account, balance of payments, and exchange rate), financial risk (bank structureand financial stability), and operative risk (legal system, attitude towards foreign investors, infrastructure, and natural conditions)(http://www.sace.it/).
11The windows are identified considering the quartiles of the M&A distribution. The observations in the control sample are chosen byrandomly drawing from each group a number of observations equal to the number of M&As in the group, so as to obtain two samplesof identical size. The last step is repeated 1000 times in order to obtain confidence intervals for the t-test statistics. The t-test statistictakes the form
zs − zc√[zs(1 − zs) + zc(1 − zc)] /Z
under H0 : zs = zc H1 : zs > zc, (3)
where Z is the total number of observations and z is the frequency of M&As. Subscripts s and c are used to highlight whether z isobserved in the original or control sample. Only the deals by firms with at least one previous M&A are considered in each period.
9
and 12.3 percent of the sample, respectively. See Table 3 for descriptive statistics.
Table 3 around here
Besides the value of the M&A (i.e., P th), the right-hand side of Equation (4) includes a vector of exogenous factors
specific to the country of origin, the target country and the country pair (i.e., ΓtH,F ) and the experience variables
as defined in Section 3. Namely, the share of firm h’s past M&As at time t in country F , third countries Z that are
geographically contiguous to F , third countries Z that are culturally contiguous to F (i.e., M th(F ), M t
h(Z contgeo F )
and M th(Z contcult F ), respectively. The shares are calculated with respect to the total number of firm h’s past
M&As at time t, since 1985. It is worth noting that our 1985-2007 data coverage yields a discrepancy between
observed and actual experience for the firms established before 1985, as we are not allowed to know whether they
already engaged in M&A operations before that date; this circumstance entails initial condition issues (see below).
The three variables are summarised in Table 3, also in terms of cross-correlation. While experience in contiguous
countries is always less than one, M th(F ) = 1 in some cases, meaning that the firm engaged in M%A deals in one
country only; these cases have been dropped in the estimation.
Table 4 reports our benchmark findings. Models (1) and (2) are used to carry out the estimation of (4) for
M&As in culturally contiguous countries; models (3) and (4) are used for M&A in geographically contiguous coun-
tries. As explained, our data do not reveal whether a firm had already engaged in M&As before 1985. We tackle
this initial condition problem by including the initial and the average value of all the experience variables (namely,
P th and the three experience variables) in a Probit random-effects estimation. Introduced by Wooldridge (2005),
this strategy enables us to deal with the fact that we have no ‘true’ values observed at the beginning of the time
window, and that we therefore cannot derive a reduced-form equation for the initial condition based on available
pre-sample information as suggested by Heckman (1981). This approach provides an alternative method to take
firm effects into account, but under the hypothesis that they are not correlated with the explanatory variables, as
seems reasonable. Estimation results are reported in Models (1) and (3)12
Table 4 around here
For vector ΓtH,F , we include the total population in the origin and the host countries. This aims to capture country
size, together with differential average years of schooling, intended as a broad measure of differences in economic
development (with positive values indicating higher economic development in the target country).
Investing in a culturally contiguous country (Models 1 and 2) is positively affected by prior experience in that
country, as measured by M th(F ). On the contrary, and notably, direct experience in the destination country is
found to be uninfluential when the target country is geographically contiguous (Models 3 and 4). This suggests
that only M&A driven by factors in the cultural sphere, as captured by our cultural distance measure, are subject
to experience effects.
This finding is confirmed by the fact that the occurrence of M&A deals in a culturally contiguous country is
also positively affected by accumulated experience in countries that are culturally contiguous to the destination
12Fixed-effects panel logit regressions have also been run for consistency; results are broadly in line with those in Table 4.
10
country, that is by M th(Z contcult F ). Conversely, M t
h(Z contgeo F ) is found to exert a negative effect, which is
evidence against the presence of experience effects associated with geographical proximity.
Interestingly, these effects tend to persist (see Model 3) in M&A operations driven by geographical contiguity
(although they disappear in Model 4), that is, in deals that we associate with transaction costs.
Regarding the control variables, higher economic development (i.e. schooling) in the target country exerts the
same, positive effect in both types of M&A deals. While positively correlated with M&As in culturally contiguous
countries, the size of the destination country seems to push in the direction when M&A in geographically contiguous
countries is considered. The value of the deal is in general not significant (also in the robustness analysis). In
summary, the above results strongly point to the presence of experience effects associated with the cultural sphere
and, notably, the absence of similar effects in association with geographical proximity. Hence, we interpret the
overall evidence as being driven by a process of learning, arguably in terms of knowledge accumulation.
5 Robustness
The robustness of the results in Table 4 is checked in several ways.
For the estimating method, we retain the Probit random-effects estimation as our preferred specification and
perform the robustness checks only with respect to this. As well as imposing the firm effects to be correlated with
our experience indicators, which seems too restrictive an assumption, the fixed-effects logit specification incurs
issues of convergence achieving in most cases.
In Table 5 we use different specifications for the control vector ΓtH,F . In Models (1) and (3), we include cross-
country differences in unit labour costs, interest rate, profit tax rate, and country risk. Results are qualitatively the
same, with experience still playing a key role, irrespective of higher tax and risk rate and the occurrence of M&A
deals in a culturally (geographically) contiguous country positively (negatively) related to ULC, which is consistent
with previous results on schooling. In Models (2) and (4), considering the data generating process described in
Appendix A, we also control for geographical distance in the cultural contiguity regression (Model 2) and cultural
distance in the geographical contiguity regression (Model 4) regression. Given the positive and significant action
exerted by geographical distance, this analysis makes the experience effects even more evident, indicating that
M&As mostly reach geographically distant markets when they are driven by cultural relationships. This evidence
explains the M&A flows occurring, for example, between the US and the UK, the US (or UK) and Australia,
and Spain/Italy/Portugal and Latin America very well. In contrast, DcultH,F acts as a deterrent with respect to
the probability of M&As in geographically (but not culturally) contiguous countries.13 It is worth noting that,
somewhat counter-intuitively, we do not include both distance measures in the same model for two reasons. First,
our dependent variable in the cultural (geographical) contiguity regression takes the value of one when cultural
(geographical) distance to the target country is below a certain threshold. Hence, cultural (geographical) distance
would absorb most of the variability in the dependent variable if included in the cultural (geographical) regression.
Second, the data generating process described in Appendix A suggests that cultural (geographical) distance is of
13A straightforward interpretation of the positive effect of geographical distance on M&As in culturally contiguous countries, andthe positive effect documented for ULC and profit tax rates might lie in the colonial heritage of the major acquiring countries (the USand UK).
11
negligible relevance in M&A deals directed toward culturally (geographically) contiguous countries. To generally
assess the concurrent impact of geographical and cultural distance in the aggregate, we perform regression analysis
in terms of total number and total value of the M&A deals taking place from H to F during the period under
consideration. Results reported in Appendix B.1, point to both types of distance exerting a negative impact.
Table 5 around here
In Table 6, we run the benchmark regressions separately for horizontal (models 1 and 2) and vertical (models 3 and
4) M&As, using a 1-digit sectoral breakdown. The analysis reveals the absence of remarkable differences between
the two types of M&As, meaning that our benchmark results can be regarded as general. Second, to confirm that
the final messages are not driven by the main players, we repeat the estimation without the first three countries in
terms of both inward and outward flows of M&As: the US, the UK, and Canada (results reported in the last two
columns). Our main results prove robust also to this check.
Table 6 around here
Finally, in Table 7, we use a different contiguity criterion for the dependent variables. In Model (1) and (2) we
redefine cultural contiguity by considering country H culturally contiguous to country F if its cultural distance
to F is lower than the value corresponding to the fifth (instead of second) percentile of country H’s distribution
of bilateral distances and, in addition, lies within the second decile of the worldwide distribution. This brings
the number of culturally contiguous country-pairs from 64 to 136 (out of 1047 in the whole database). Results
are unaffected. In Model (3) and (4) we change the definition of geographical contiguity from sharing a common
border to a definition that is consistent with the definition of cultural contiguity used in benchmark regressions.
Country H is considered geographically contiguous to country F if its geographical distance to F is lower than the
value corresponding to the second percentile of country H’s distribution of bilateral distances and, in addition, lies
within the second decile of the worldwide distribution. Benchmark results are confirmed in this case.
Table 7 around here
6 Conclusions
We investigated the eventual presence of knowledge-related experience effects and revealed cross-patterns of (country-
specific) cultural distance and (firm-specific) experience.
In particular, we exploited the cross-time and cross-space correlation in a cross-border M&A database (drawn
from the Thomson Financial Security Database) with worldwide coverage. We obtained empirical evidence on
whether a firm’s history in terms of prior M&As contributes to generating experience effects, associated with
knowledge accumulation, exerting a positive impact on the probability to invest in the same country and/or in
culturally similar countries again.
12
Our empirical strategy is built on the hypothesis that, while knowledge can be accumulated, the burden of
transport costs related to economic activity in a country cannot. Accordingly, experience effects are expected to be
at play in M&A in culturally contiguous countries (arguably driven by knowledge costs) and not at play in M&A
deals in geographically contiguous countries (arguably driven by spatial costs).
The analysis is conducted on a bilateral measure of cultural distance obtained as a weighted average of linguistic,
religious, and genetic distance indicators. The measure is conceived to capture the idea of the cultural heritage
originated by historic linkages, following Guiso et al. (2006) and Spolaore and Wacziarg (2016). We calculate
linguistic and religious distances based on information drawn from Ethnologue and the CIA World Factbook,
respectively. The linguistic index is more articulated than comparable available measures (e.g. Spolaore and
Wacziarg, 2016; Melitz and Toubal, 2014): 195 countries are covered (against 156 in Spolaore and Wacziarg, 2016);
bilateral distance is always maximum or always minimum in no cases (against 35 out of 195 in Melitz and Toubal,
2014).
Econometric results strongly support the hypothesis that firms’ M&A choices are significantly influenced by
experience effects related to knowledge accumulation. However, our approach highlights that to uncover such
experience effects, we need to look where they potentially are at play, that is, in knowledge-intensive deals/activities.
From a policy perspective, our results highlight how, by compressing firms’ involvement in FDI activities,
back-shoring policies are likely to disrupt the process of knowledge accumulation. This introduces an additional
channel through which protectionism can reduce the welfare gains associated with market openness. While this is a
potential research line along which our results might be developed, it would also be interesting investigating on the
potential asymmetric experience effects of cultural and geographical contiguity, as long as distance is recognized to
be not reciprocal (that is, when distance from A to B is not equivalent to distance from B to A), as the analysis
by Boschma et al. (2016) seems to suggest.
13
References
Ahern, K. R., Daminelli, D., & Fracassi, C. (2015). Lost in translation? The effect of cultural values on mergers
around the world. Journal of Financial Economics, 117(1), 165-189.
Ambos, B., & Hakanson, L. (2014). The concept of distance in international management research. Journal of
International Management, 20(1), 1-7.
Barkema, H. G., Bell, J. H., & Pennings, J. M. (1996). Foreign entry, cultural barriers, and learning. Strategic
management Journal, 17(2), 151-166.
Barro, R. & Lee, J.W. (2013). A new data set of educational attainment in the world, 1950-2010, rev.2 first version
on NBER Working Papers, 15902.
Baum, J. A. C., Li, S., & Usher, J. (2000) Making the next move: how experiential and vicarious learning shape
the locations of chains’ acquisitions. Administrative Science Quarterly, 45, 766-801.
Benito, G. R., & Gripsrud, G. (1992). The expansion of foreign direct investments: discrete rational location
choices or a cultural learning process? Journal of International Business Studies, 23(3), 461-476.
Blonigen, B. A., Davies, R. B., Waddell, G. R., & Naughton, H. T. (2007). FDI in space: Spatial autoregressive
relationships in foreign direct investment. European Economic Review, 51(5), 1303-1325.
Boschma, R., Marrocu, E., & Paci, R. (2016). Symmetric and asymmetric effects of proximities. The case of M&A
deals in Italy. Journal of Economic Geography, 16(2), 505-535.
Brakman S., Garretsen H., & van Marrewijk, C. (2006). Cross border Mergers and Acquisitions: the facts as a
guide for international economics, CESifo Working Paper, 1823.
Buckley, P. J., & Casson, M. (1976). A long-run theory of the multinational enterprise. In The future of the
multinational enterprise. Palgrave Macmillan, London.
Camagni, R. (1991) Local ’milieu’, uncertainty and innovation networks: Towards a new dynamic theory of eco-
nomic space, in: R. Camagni (Ed.) Innovation Networks, 121-144, London, Belhaven Press.
Capello, R. (1999) SME clustering and factor productivity: A milieu production function model, European Planning
Studies, 7, 719-735.
Carlson, S. (1974). International transmission of information and the business firm. The Annals of the American
Academy of Political and Social Science, 412(1), 55-63.
Chao, M.C.-H., & Kumar, V. (2010). The impact of institutional distance on the international diversity-performance
relationship. Journal of World Business, 45(1), 93-103.
Chapman, M. A., Clegg, H. G. M., & Buckley, J. P. J. (2008). Close neighbors and distant friends - perceptions of
cultural distance. International Business Review, 17, 217-234.
Coeurdacier, N., R. A. De Santis & Aviat, A. (2009). Cross-Border Mergers and Acquisitions and European
Integration, Economic Policy, 24(57), 55-106.
Cyert, R. M., & March, J. G. (1963). A behavioral theory of the firm. NJ: Englewood Cliffs.
14
Davidson, W. H. (1980). The location of foreign direct investment activity: Country characteristics and experience
effects. Journal of international business studies, 11(2), 9-22.
Delios, A., & Beamish, P. W. (2001). Survival and profitability: The roles of experience and intangible assets in
foreign subsidiary performance. Academy of Management Journal, 44(5), 1028-1038.
Di Giovanni, J. (2005). What Drives Capital Flows? The Case of Cross-Border M&A Activity and Financial
Deepening, Journal of International Economics, 65(1), 127-49.
Di Guardo, M. C., Marrocu, E., & Paci, R. (2016). The concurrent impact of cultural, political, and spatial
distances on international mergers and acquisitions. The World Economy, 39(6), 824-852.
Dunning J. H. (1977). Trade, location of economic activity and the multinational enterprise: a search for an eclectic
approach, in The International Allocation of Economic Activity, Ohlin B., Hesselborn P. O., & Wijkman P. M. Eds.,
Macmillan, London, 395-417.
Fearon, J.D. (2003). Ethnic and cultural diversity by country*. Journal of Economic Growth, 8(2), 195-222.
Gaur, A. S., Delios, A., & Singh, K. (2007). Institutional environments, staffing strategies and subsidiary perfor-
mance. Journal of Management, 33(4), 611-636.
Guillen, M.F. (2003). Experience, imitation, and the seuence of foreign entry: wholly owned and joint-venture
manufacturing by South Korean firms and business groups in China, 1987-1995, Journal of International Business
Studies, 34(2), 185-98.
Guiso, L., Sapienza, P., and Zingales, L. (2006). Does Culture Affect Economic Outcomes? Journal of Economic
Perspectives, 20(2), 23-48.
Haleblian, J., & Finkelstein, S. (1999). The influence of organizational acquisition experience on acquisition per-
formance: A behavioral learning perspective. Administrative Science Quarterly, 44, 29-56.
Heckman, J.J. (1981). The Incidental Parameters Problem and the Problem of Initial Conditions in Estimating a
Discrete Time - Discrete Data Stochastic Process, in Structural Analysis of Discrete Panel Data with Econometric
Applications, Manski and D. McFadden (eds.), Cambridge: MIT Press.
Hijzen, A., Gorg, H., & Manchin, M. (2008). Cross-border mergers and acquisitions and the role of trade costs.
European Economic Review, 52(5), 849-866.
Hofstede, G. (1980). Culture’s Consequences: International Differences in Work Related Values, Beverly Hills, CA,
Sage Publications.
Hutzschenreuter, T., Voll, J., & Verbeke, A. (2011). The impact of added cultural distance and cultural diversity
on international expansion patterns: A Penrosean perspective. Journal of Management Studies, 48(2), 305-329.
Hyun, H. J. & Kim H. H. (2010). The Determinants of Cross-Border M&As: The Role of Institutions and Financial
Development in the Gravity Model, The World Economy, 33(2), 292-310.
Johanson, J., & Vahlne, J. E. (1977). The internationalization process of the firm-a model of knowledge development
and increasing foreign market commitments. Journal of international business studies, 8(1), 23-32.
15
Kaasa, A., M. Vadi & Varblane, U. (2013). European Social Survey as a Source of New Cultural Dimensions
Estimates for Regions, International Journal of Cross Cultural Management, 13(2), 137-57.
Kogut, B. & Zander, U. (1993). Knowledge of the firm and the evolutionary theory of the multinational corporation,
Journal of International Business Studies, 24, 625-45.
Kogut, B., & Singh, H. (1988). The effect of national culture on the choice of entry mode. Journal of International
Business Studies, 19(3), 411-432.
Lawson, C. (2000) Collective learning, system competences and epistemically significant moments, in: D. Keeble and
F. Wilkinson (Eds) High-technology Clusters, Networking and Collective Learning, 182-198 (Aldershot: Ashgate).
Leung, K., Bhagat, R., Buchan, N., Erez, M., & Gibson, C. (2005). Culture and international business: Recent
advances and their implications for future research. Journal of International Business Studies, 36, 357-378.
Levitt, B., & March, J. G. (1988). Organizational learning. Annual Review of Sociology, 14, 319-340.
Lundvall, B.A. (Ed.) (1992) National Systems of Innovation: Towards a Theory of Innovation and Interactive
Learning, London: Pinter.
Luo, Y., & Shenkar, O. (2011). Toward a perspective of cultural friction in international business. Journal of
International Management, 17(1), 1-14.
McCann P (1998) The economics of industrial location: A logistics-costs approach. Springer, Berlin Heidelberg
New York
McCann, P., & Shefer, D. (2004). Location, agglomeration and infrastructure. Papers in regional science, 83(1),
177-196.
Melitz J. & Toubal F. (2014). Native Language, Spoken Language, Translation and Trade. Journal of International
Economics, 92(2), 2014 , p. 351-363.
Mitra, D., F. Golder, P. N. (2002). Whose culture matters? Near-market knowledge and its impact on foreign
market entry timing. Journal of Marketing Research, 39(3), 350-365.
Perkins, S. E. (2014). When does prior experience pay? Institutional experience and the multinational corporation.
Administrative Science Quarterly, 59(1), 145-181.
Popli, M., Akbar, M., Kumar, V., & Gaur, A. (2016). Reconceptualizing cultural distance: The role of cultural
experience reserve in cross-border acquisitions. Journal of World Business, 51(3), 404-412.
Santangelo, G. D., & Stucchi, T. (2018). Internationalization through exaptation: The role of domestic geographical
dispersion in the internationalization process. Journal of International Business Studies, 49(6), 753-760.
Shenkar, O. (2012). Beyond cultural distance: Switching to a friction lens in the study of cultural differences.
Journal of International Business Studies, 43(1), 12-17.
Spolaore, E., & Wacziarg, R. (2009). The Diffusion of Devolpment. Quarterly Journal of Economics, 124(2),
469-529.
16
Spolaore, E., & Wacziarg, R. (2016). Ancestry, Language and Culture. In Ginsburgh V. and Weber S. (eds.), The
Palgrave Handbook of Economics and Language, Ch. 7, 174-211, London: Palgrave Macmillan.
Storper, M. (1995) The resurgence of regional economies, ten years later: The region as a nexus of untraded
interdependencies, European Urban and Regional Studies, 2, 191-221.
Storper, M. (1997) The Regional World, New York: Guilford Press.
Todtling, F., Lehner, P., & Trippl, M. (2006). Innovation in knowledge intensive industries: The nature and
geography of knowledge links. European planning studies, 14(8), 1035-1058.
Very, P., Lubatkin, M., Calori, R., & Veiga, J. (1997). Relative standing and the performance of recently acquired
European firms. Strategic Management Journal, 18(8), 593-614.
Wooldridge, J.M. (2005). Simple solutions to the initial conditions problem in dynamic, nonlinear panel-data
models with unobserved heterogeneity. Journal of Applied Econometrics, 20, 39-54.
Yildiz, H. E. (2014). Not all differences are the same: Dual roles of status and cultural distance in sociocultural
integration in cross-border M&As. Journal of International Management, 20(1), 25-37.
17
Table 1: M&A data: descriptive statistics by country of origin.
COUNTRY Abbr. N Value N (share of tot.) Value (share of tot.)United States US 6474 1112325 26.53% 20.40%United Kingdom UK 5869 1269393 24.05% 23.28%Canada CA 2409 314659 9.87% 5.77%France FR 1102 557689 4.52% 10.23%Germany WG 1060 485834 4.34% 8.91%Australia AU 1046 175792 4.29% 3.22%Netherlands NT 871 331818 3.57% 6.09%Sweden SW 830 133608 3.40% 2.45%Japan JP 751 110867 3.08% 2.03%Ireland IR 581 44217 2.38% 0.81%Italy IT 504 120538 2.07% 2.21%Spain SP 497 229883 2.04% 4.22%Switzerland SZ 493 249992 2.02% 4.59%Norway NO 405 50391 1.66% 0.92%Finland FN 342 54216 1.40% 0.99%Belgium BL 338 88680 1.39% 1.63%Denmark DN 324 49160 1.33% 0.90%Austria AS 201 29538 0.82% 0.54%New Zealand NZ 139 16527 0.57% 0.30%Portugal PO 88 14168 0.36% 0.26%Greece GR 78 12436 0.32% 0.23%TOTAL - 24402 5451731 100% 100%
Table 2: T-test statistics.
# obs in # obs in ratio* t-test* P5* P95*Sample Control*
(1) F 4862 3658 1.329 15.568 15.106 16.056
(2) F contgeo H 1073 894 1.200 4.170 3.841 4.498
(3) F contcult H 2221 1844 1.204 6.377 5.974 6.784
(4) F cont2geo H 681 564 1.208 3.385 2.337 4.380
(5) F cont2cult H 935 1003 0.933 -1.601 -2.670 -0.511
* Bootstrapped values (average values after 1000 replications)
T-test on the sample frequency of M&As towards:(1) countries in which the firm has already invested at least once;(2) geographically contiguous countries in which the firm has already invested at least once;(3) culturally contiguous countries in which the firm has already invested at least once;(4) second-order geographically contiguous countries in which the firm has already invested at least once;(5) second-order culturally contiguous countries in which the firm has already invested at least once.
H: acquisition country; F : target country; Z: third country.
18
Table 3: Experience variables: correlation table.
Statistics: mth(F contcult H) mt
h(F contgeo H) Mth(F ) Mt
h(Z contcult F ) Mth(Z contgeo F )
Min 0 0 0 0 0Max 1 1 1 0.9 0.9Mean 0.269 0.123 0.144 0.060 0.041Std.Dev. 0.443 0.328 0.237 0.146 0.122
Correlations:
mth(F contcult H) 1
mth(F contgeo H) -0.227 1
Mth(F ) 0.208 0.123 1
Mth(Z contcult F ) 0.061 -0.013 -0.092 1
Mth(Z contgeo F ) -0.112 -0.082 -0.102 0.242 1
Table 4: Experience effects: benchmark estimates.
(1) (3)
mth(F contcult H) mt
h(F contgeo H)
P th -0.005* 0.002
(-2.45) (1.30)
Mth(F ) 0.251*** -0.008
(13.28) (-0.62)
Mth(Z contgeo F ) -0.329*** -0.059*
(-9.36) (-2.41)
Mth(Z contcult F ) 0.160*** 0.068***
(7.10) (4.55)
Population H 0.000*** 0.000**(13.34) (2.60)
Population F 0.000*** -0.000***(10.07) (-14.61)
Schooling (F-H) 0.061*** 0.012***(34.87) (9.52)
N 11953 11953Converged 1 1
Estimation method: Random-effects Probit with average and initial values of time-varying regressors included
Firm and time effects included in all specifications.
Marginal effects reported. Standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001.
Legend:H: acquisition country; F : target country; Z: third country.
Dependent variable:
mth(F contcult H) = 1 if countries H and F are culturally contiguous and not geographically contiguous
mth(F contgeo H) = 1 if countries H and F are geographically contiguous and not culturally contiguous
Regressors:P t
h = acquisition price;Mt
h(F ) = share of firm h’s past M&As in country F ;Mt
h(Z contgeo F ) = share of firm h’s past M&As in geographically contiguous countries to F
Mth(Z contcult F ) = share of firm h’s past M&As in culturally contiguous countries to F ;
(shares calculated with respect to the total number of firm h’s past M&As at time t):
19
Tab
le5:
Rob
ust
nes
s:d
iffer
ent
mod
elsp
ecifi
cati
on
s.
(1)
(2)
(3)
(4)
mt h
(Fcontc
ultH
)m
t h(F
contc
ultH
)m
t h(F
contg
eoH
)m
t h(F
contg
eoH
)
Pt h
-0.0
10***
-0.0
08***
0.0
01
0.0
01
(-4.8
7)
(-4.1
8)
(0.6
7)
(0.9
8)
[1em
]M
t h(F
)0.1
50***
0.1
95***
0.0
09
-0.0
01
(7.7
7)
(10.2
0)
(0.6
4)
(-0.0
5)
Mt h(Z
contg
eoF
)-0
.167***
-0.1
30***
-0.0
80**
-0.0
71*
(-4.4
2)
(-3.6
4)
(-2.7
1)
(-2.4
6)
Mt h(Z
contc
ultF
)0.1
09***
0.0
91***
0.0
93***
0.0
84***
(5.1
3)
(3.9
3)
(5.9
5)
(5.4
1)
Schooling(F
-H)
0.0
47***
0.0
45***
0.0
08***
0.0
06***
(20.4
9)
(20.0
7)
(5.1
9)
(3.7
0)
Population
H0.0
00***
0.0
00***
0.0
00**
0.0
00**
(33.3
9)
(21.4
2)
(2.8
9)
(2.9
7)
Population
F0.0
00***
0.0
00***
-0.0
00***
-0.0
00***
(42.7
5)
(43.6
8)
(-10.4
5)
(-6.4
9)
ULC
(F-H
)0.0
00***
0.0
00***
-0.0
00***
-0.0
00***
(13.5
1)
(14.6
7)
(-5.3
9)
(-5.6
2)
Intrate
(F-H
)-0
.000
-0.0
01
-0.0
16***
-0.0
16***
(-0.0
4)
(-0.8
5)
(-8.5
2)
(-8.7
0)
Tax(F
-H)
0.0
12***
0.0
08***
0.0
01**
0.0
01
(22.0
3)
(13.7
3)
(2.6
1)
(1.8
7)
Risk(F
-H)
-0.0
27***
-0.0
43***
-0.0
07**
-0.0
04
(-5.6
5)
(-9.1
9)
(-2.6
5)
(-1.2
9)
Dgeo
H,F
0.4
94***
(28.7
6)
Dcult
H,F
-0.1
07***
(-5.6
3)
N9843
9843
9843
9843
Converg
ed
11
11
Est
imati
on
meth
od:
Random
-eff
ects
Pro
bit
wit
havera
ge
and
init
ial
valu
es
of
tim
e-v
ary
ing
regre
ssors
inclu
ded.
Fir
mand
tim
eeff
ects
inclu
ded
inall
specifi
cati
ons.
Marg
inal
eff
ects
rep
ort
ed.
Sta
ndard
err
ors
inpare
nth
ese
s.*p<
0.0
5,
**p<
0.0
1,
***p<
0.0
01.
Legend:se
eTable4.
20
Tab
le6:
Rob
ust
nes
sch
ecks:
esti
mati
on
on
sub-s
am
ple
s.
(1)
(2)
(3)
(4)
(5)
(6)
mt h
(Fcontc
ultH
)m
t h(F
contg
eoH
)m
t h(F
contc
ultH
)m
t h(F
contg
eoH
)m
t h(F
contc
ultH
)m
t h(F
contg
eoH
)
Pt h
-0.0
08***
0.0
02
-0.0
00
0.0
01
-0.0
05**
0.0
06
(-3.4
9)
(1.3
9)
(-0.1
2)
(.)
(-3.1
6)
(.)
Mt h(F
)0.2
82***
0.0
00
0.1
96***
-0.0
30
0.0
54*
-0.0
44
(12.9
7)
(0.0
2)
(5.5
3)
(.)
(2.1
8)
(.)
Mt h(Z
contg
eoF
)-0
.335***
-0.0
60*
-0.3
27***
-0.0
60
-0.1
66***
-0.0
09
(-7.9
2)
(-2.1
4)
(-4.9
7)
(.)
(-5.1
4)
(.)
Mt h(Z
contc
ultF
)0.1
72***
0.0
77***
0.1
30**
0.0
47
0.2
27***
-0.0
12
(6.6
4)
(4.4
0)
(2.7
2)
(.)
(11.0
9)
(.)
Population
H0.0
00***
0.0
00*
0.0
00***
0.0
00
0.0
00
-0.0
00
(9.9
9)
(2.3
3)
(9.9
7)
(.)
(0.0
1)
(.)
Population
F0.0
00***
-0.0
00***
0.0
00***
-0.0
00
-0.0
00
-0.0
00
(7.5
0)
(-12.7
5)
(6.8
6)
(.)
(-0.0
1)
(.)
Schooling(F
-H)
0.0
59***
0.0
11***
0.0
64***
0.0
13
0.0
37***
0.0
01
(29.4
7)
(7.7
8)
(18.7
0)
(.)
(15.6
4)
(.)
N8834
8834
3119
3103
7285
7272
Converg
ed
11
11
11
Est
imati
on
meth
od:
Random
-eff
ects
Pro
bit
wit
havera
ge
and
init
ial
valu
es
of
tim
e-v
ary
ing
regre
ssors
inclu
ded.
Fir
mand
tim
eeff
ects
inclu
ded
inall
specifi
cati
ons.
Marg
inal
eff
ects
rep
ort
ed.
Sta
ndard
err
ors
inpare
nth
ese
s.*p<
0.0
5,
**p<
0.0
1,
***p<
0.0
01.
Legend:se
eTable4.
21
Tab
le7:
Rob
ust
nes
sch
ecks:
diff
eren
tco
nti
gu
ity
crit
eria
.
(1)
(2)
(3)
(4)
mt h
(Fcontc
ultH
)m
t h(F
contc
ultH
)m
t h(F
contg
eoH
)m
t h(F
contg
eoH
)
Pt h
-0.0
06**
0.0
02
-0.0
02
0.0
00
(0.0
02)
(0.0
04)
(0.0
02)
(0.0
04)
Mt h(F
)0.2
62***
0.3
51***
0.0
55*
0.1
56
(0.0
24)
(0.0
98)
(0.0
27)
(0.0
83)
Mt h(Z
contg
eoF
)-0
.446***
-0.5
04***
-0.0
43*
-0.1
01
(0.0
46)
(0.1
48)
(0.0
21)
(0.0
60)
Mt h(Z
contc
ultF
)0.2
61***
0.2
88***
0.0
69**
0.1
07
(0.0
21)
(0.0
80)
(0.0
21)
(0.0
59)
Population
H0.0
00***
0.0
00
-0.0
00***
-0.0
00
(0.0
00)
(0.0
00)
(0.0
00)
(0.0
00)
Population
F0.0
00***
0.0
00***
-0.0
00***
-0.0
00**
(0.0
00)
(0.0
00)
(0.0
00)
(0.0
00)
Schooling(F
-H)
0.0
56***
0.0
83***
0.0
02
0.0
03
(0.0
02)
(0.0
22)
(0.0
02)
(0.0
04)
N7756
4255
11921
5749
Pse
udoR
20.3
63
0.0
47
Converg
ed
11
11
Est
imati
on
meth
od:
Random
-eff
ects
Pro
bit
wit
havera
ge
and
init
ial
valu
es
of
tim
e-v
ary
ing
regre
ssors
inclu
ded.
Fir
mand
tim
eeff
ects
inclu
ded
inall
specifi
cati
ons.
Marg
inal
eff
ects
rep
ort
ed.
t-st
ati
stic
sin
pare
nth
ese
s.*p<
0.0
5,
**p<
0.0
1,
***p<
0.0
01.
Legend:se
eTable4.
22
Geographical and Cultural Patterns in Cross-Border
Mergers and Acquisitions: The Role of Experience
APPENDIX
Massimo Del Gatto1
‘G.d’Annunzio’ University, LUISS Guido Carli, CRENoS
Carlo S. MastinuUniversity of Cagliari
1Corresponding author. Contact details: Massimo Del Gatto; ‘G.d’Annunzio’ University, Department of Economics (DEc); VialePindaro, 42; 65127 - PESCARA (ITALY); [email protected]
1
A Appendix: Data generating process (an empirical model)
According to the literature discussed in Section 1, past investment in a given location is recognized to generate
experience effects insofar it is related to knowledge, along its various dimensions. In turn, the idea of knowledge
interrelates the ‘cultural’ sphere in different forms and, as a results, the effect exerted by cultural differences is
likely to be different from what can be measured in terms of cultural distance scores between the home and host
countries. it is in fact argued that, by investing in a given country, the firm increases its capacity to take advantage
of further investment in that country or in a country which is in turn culturally “similar” to that one.
We hereby sketch out a very simple model which consistent with this idea. This enables us to highlight the
data generating process underlying our testing hypothesis.
Firms engage in M&As to pursue relationship-specific activities. Acquiring firms benefit from M&As if and
only if the target firm fulfills given requirements. The acquisition price P th is a sufficient statistic for the set of
requirements that firm h sets itself, at time t. Only potential acquisitions of price P th are considered by firm h; P t
h
is the same, irrespective of the country in which the target firm is located.
Using lowercase letters to refer to firms and uppercase letters to refer to countries, the expected net profit
Πth,f (·), associated with the acquisition of firm f (in country F ) by firm h (in country H) at time t, can be
expressed as
Et[Πt
h,f
]= Et
[Rt
h,f (Γ, ·)− Sth,f (Dcult
H,F ,Φth,F )
]− T t
h,f (DgeoH,F )− P t
h (1)
where Rth,f denotes revenues, T t
h,f refers to spatial (i.e, transport) costs and Sth,f to knowledge costs, with
∂Sth,f
∂DcultH,F
> 0
and∂St
h,f
∂Φth,F
> 0. DgeoH,F and Dcult
H,F indicate geographic and cultural distance between country H and country F ,
respectively.
While spatial costs are known at time t, knowledge costs, as they also depend on experience, through the term
Φth,F = Φ(M t
h(F )
, Mth(Z contcult F ))(2)is assumed to be a function of the cumulative distribution function (i.e., number of past
M&As), at time t, of firm h’s M&As in country F - i.e., M th(F ) - and in countries Z that are culturally contiguous
to F - i.e., Mth(Z contcult F ).
Γ is a vector of country characteristics, specific to H and F (e.g., factor costs, taxation, etc.), where the dot
highlights that the general formulation in (1) does not aim at specifying how firms’ revenues are influenced by other
firm-specific factors (e.g., presence of managers with a particular degree of knowledge of a given market and total
factor productivity).
To keep the model as simple as possible, we assume that the search activity of the firm always produces a match
with a potential target firm priced P th (hence, with a firm that fully meets firm h’s requirements). Only one match
is produced and evaluated at each period and the decision concerning whether to take (i.e., mth,f = 1) or leave
(i.e., mth,f = 0) the opportunity is made independently of the decision to invest in other countries in the same
period. Production adjusts freely to current market conditions, entailing the profit-maximizing production level to
be always chosen.
2
In each period t, firm h chooses the infinite sequence mt[+]h,F that maximizes the expected present value of net
profits. The maximized payoff takes the form
V th,f (Ωt
h) = maxm
t[+]h,F
Et
∞∑j=t
δj−tΠth,f (·) | Ωt
h
(3)
where δ is the one-period discount rate and Ωth is firm h’s information set at time t.
Using Bellman’s equation, firm h’s current decision is the value of mth,F that satisfies
V th,f (Ωt
h) = maxmt
h,F
Πth,f (·) + δEt
[V t+1h,f (Ωt
h) |M t+1h,F ,Mh,F t+1
]. (4)
The firm will decide to take the M&A opportunity (that is, mth,F = 1) if
δ(Et[V t+1h,f (Ωt
h) | mth,F = 1
]− Et
[V t+1h,f (Ωt
h) | mth,F = 0
])+ Et
[Rt
h,f − Sth,f
]− T t
h,f > P th (5)
As well as to the acquisition target that the firm sets itself (i.e., P th), Equation (5) relates a firm’s decision to
engage in an M&A to geographical distance, cultural distance and experience.2
Now, consider two activities i and j, identical in everything but the fact that the former is ‘intensive’ in spatial
costs, so that[∂Πt
h,f
∂T th,f
]i>[∂Πt
h,f
∂T th,f
]j
while the latter is intensive in knowledge costs, so that so that[∂Πt
h,f
∂Sth,f
]j>[
∂Πth,f
∂Sth,f
]i. As long as geographical (cultural) contiguity minimizes the cost of geographic (cultural) distance, the
probability to engage in transport intensive M&As is the highest in geographically contiguous countries (where T th,f
is minimum), while the probability of engaging in knowledge intensive M&As is highest in the countries featuring
the best combination between cultural distance and experience; that is, starting from the initial condition in which
both M th(Z) = 0 and M t
h(Z contcult F ) = 0 ∀Z, in culturally contiguous countries (such as Sth(Z) is minimized).
Such a data generating process is consistent the Testable Hypothesis stated in Section 2. In fact, M&A occurring
in culturally contiguous are arguably associated to a knowledge cost-minimizing choice and, as such, are expected
to be positively associated to M th(Z) and M t
h(Z contcult F ) and to not depend on DcultH,F . There is no a priori
expectation on the effect of DgeoH,F , as long as (being the activity knowledge intensive), the impact of transport
costs is comparatively marginal and it might be reasonable to choose geographically distance but culturally similar
countries (think about former UK colonies for investment from the UK) in horizontal M&A a la Barba Navaretti and
Venables, 2004. Differently, M&As in geographically contiguous countries are expected to take place independently
of experience, to not depend on DgeoH,F (which is already minimized by contiguity) and to still depend on cultural
distance, eventually.
2Note that with Sthf (·) + T t
h,f (·) + P th > Et[Rt
h,f (·)], firm h still chooses mth,F = 1 if the expected contribution to future profits
exceeds the negative expectation for the current period. That might be the case for a firm that wants to invest in a given country inorder to improve its production abilities through learning.
3
B Appendix: M&A Data
We use M&A data drawn from the Thomson Financial Security Database. The database has world coverage and,
for each deal, reports information on country of origin and country of destination, year, data of announcement,
value of the acquisition, ISIC industry of the acquiring and the target firms.
While a detailed description of the database can be found in Brakman et al. (2006), we hereby report descriptive
statistics and visual inspection along some key dimensions for the analysis carried out in the paper.
As known, M&As, as FDI in general, come in ‘waves’. Such circumstance is highlighted by Figure B.1 using the
FDI Database provided by UNCTAD. The Figure also underlines the growing importance of the M&A flows, as a
subset of FDI, from the 90s to the 2008 economic crisis, since when the share of FDI occurring in form of M&As
progressively shrinks.
Figure B.2 highlights how the time variability in the Thomson Financial Security Database, in terms of number
and value of M&As deals is broadly consistent with the UNCTAD Data. Figure B.3 shows the country patterns
(in terms of number of M&As). There are no substantial differences across countries over time.
From a static perspective, Figure B.4 describes the country breakdown fo the data, highlighting how the biggest
players in terms of outward M&As tend also to be the biggest players in terms of inward flows.
M&As occur mostly within a narrow group of developed countries, among which the US is the undisputed leader
and, as highlighted by Figure B.5, the vast majority of M&As is directed toward OECD countries.
Figure B.6 shows that the industries more involved in M&As deals are those in the the manufacturing sector,
followed by finance, insurance, real estate, and services industries.
Figures B.7, B.8 and B.9 provide a geographical representation of the data for three key cases: US, Spain,
Japan. A considerable share of the M&A flows originated in US and Spain is directed toward South America. In
the case of the US, this circumstance can be usefully explained in terms of geographical proximity. In the case of
Spain, on the contrary, the large amount of M&A in such a geographically distant area can be explained by the
cultural proximity related to the Spanish colonial heritage. Indeed, Japan displays very limited interests in that
area. This offers a first intuition of the potential role of cultural distance.
B.1 Aggregate correlation with geographical and cultural distance.
In this Section we address the concurrent impact of geographical and cultural distance on aggregate M&A flows
from country H to country F , occurred in the period 1985-2007, in terms of: i) total number of deals (i.e., models
1 to 3 in Table A1); ii) total value of the deals (i.e., models 4 to 6 in Table A1).
In the first case, the estimated equation is
M tH,F = δDi
H,F + γΓH,F + ε with i = geo, cult. (6)
where DH,F is a vector encompassing geographic (DgeoH,F ) and cultural (Dcult
H,F ) distance; ΓH,F is a vector of controls
specific to the host country and to the country of origin; ε is the iid error.
Since we observe only the deals that effectively took place within the period, the analysis poses a zero inflation
4
problem.3 To deal with this issue, Equation (6) is estimated using a zero-inflated negative binomial regression.
In particular, a hurdle model is considered, in which the equation for the first step (the equation that determines
whether the observed count is zero) includes the same variables used in the second step.
The estimation results are reported in Table A1 (models 1 to 3). We use our benchmark (parsimonious)
regression in which only total population and differential schooling are as country controls. In model (1) the full
sampe is used, while models (2) and (3) restrict the analysis to vertical and horizontal M&As, respectively.
Both geographic and cultural distance from the target country are significant determinants of both the count
(i.e., first stage) and the zero inflation (i.e., second stage) process. The sign is as expected: higher distance is
associated with more zeros, in the first stage, and a lower number of deals, in the second stage. However, the
cultural distance effect is larger in magnitude.
Regarding the control variables, M&A flows are mostly directed toward large (Population) countries, which is
expected. As for differential Schooling (proxy for differences in terms of degree of economic development) between
the host and the origin country, while less developed countries tend to be more targeted in the first stage, economic
development exerts a positive impact on the number of deals in the second stage (count process). While leaving
the sign of the coefficients unaffected, disentangling between horizontal and vertical M&As only (slightly) affect
their magnitude.
Models (4), (5) and (6) perform a similar regressions but, instead of the total number of M&A deals, their
aggregate value is considered. The estimating equation is similar to 6; however, the regressors are now expressed in
logarithms, in order to address the zero-inflation issue through the Poisson pseudo-maximum likelihood estimator
(PPML) introduced by Silva and Tenreyro (2006):
PTH,F = δ ln DH,F + γ ln ΓF + ZH + εt. (7)
PTH,F is the aggregate value of the M&A deals from country H to country F .
Regression results are broadly consistent with those produced by the second stage (i.e., count process) of the
zero inflated binomial regression approach. Also in this case, disentangling between horzintal and vertical deals
does not affect the final message of the regressions.
3When the dataset is filled in considering all possible combinations of countries and sectors in each year, we end up with 254100observations, 250544 of which are zeros.
5
31
Fonte: Baldwin e Martin (1999) su dati Dunning (1983) e Unctad - Fdi database.
Figura 34: Distribuzione Geografica degli Fdi, 1914-1996.
Fonte: elaborazione su dati Unctad - Fdi database.
Figura 35: Flussi mondiali (in valore) di Fdi e M&A.
0
200000
400000
600000
800000
1000000
1200000
1400000
1600000
1800000
2000000
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
FDI M&A
Figure B.1: FDI and M&A trends (in value).
Source: our calculation on UNCTAD - FDI Database
020
0000
4000
0060
0000
8000
00Va
lue
050
010
0015
0020
00N
umbe
r
1985 2000 2007.
Number Value
Figure B.2: Patterns of M&A: number of deals by year.
6
123
415
4
154
822
2
132
158
512
0
110
174
367
811
1
145
594
120
112
266
397
151
2546
4
1762
0
413
2
1985 2007 1985 2007 1985 2007 1985 2007
1985 2007
AS AU BL CA DN
FN FR GR IR IT
JP NO NT NZ PO
SP SW SZ UK US
WG
M&A
(num
ber)
.
Figure B.3: Patterns of M&As: number of deals by acquiring country.
France
Canada
Germany
Australia
Netherlands
Others
US (26.6%)UK (24%)
Canada
Germany
Australia
France
Nether lands
Others
UK (15.6) US (27.4)
Distribution by acquiring country Distribution by target country
Figure B.4: M&A data: number of M&As by acquiring and target country.
7
0 20 40 60 80 100
20072006200520042003200220012000199919981997199619951994199319921991199019891988198719861985
%
OECD Non-OECD
Target countries: OECD Vs non-OECD
Figure B.5: M&A data: distribution by OECD membership of the target country.
0 5,000 10,000 15,000 20,000
Agriculture, Forestry & Fishing
Communications
Construction
Finance, Insurance & Real Estate
Manufacturing
Mining
Public Administration
Retail Trade
Services
Transportation & storage
Wholesale Trade
number value
Figure B.6: M&A data: distribution by sector.
8
Figure B.7: M&A data: world map of US outward MAs (number).
Figure B.8: M&A data: world map of Spanish outward MAs (number).
Figure B.9: M&A data: world map of Japanese outward MAs (number).
9
Tab
leA
1:C
oncu
rren
tim
pact
of
geo
gra
ph
ical
an
dcu
ltu
ral
dis
tan
ce:
aggr
egate
regre
ssio
ns.
(1)
(2)
(3)
(4)
(5)
(6)
Dgeo
H,F
-1.2
64***
-1.6
91***
-1.3
25***
-0.5
86***
-0.4
78***
-0.6
22***
(-0.1
22)
(-0.1
50)
(-0.1
64)
(-0.1
05)
(-0.1
06)
(-0.1
23)
Dcult
H,F
-3.7
64***
-3.8
18***
-4.2
69***
-1.3
15***
-1.5
16***
-1.2
37***
(-0.1
84)
(-0.2
12)
(-0.2
19)
(-0.1
49)
(-0.1
53)
(-0.1
88)
Population
H0.0
00***
0.0
00***
0.0
00***
1.0
65***
1.1
40***
1.0
44***
(0.0
00)
(0.0
00)
(0.0
00)
(-0.1
70)
(-0.1
31)
(-0.2
01)
Population
F0.0
00***
0.0
00***
0.0
00***
1.1
65***
1.0
40***
1.2
05***
(0.0
00)
(0.0
00)
(0.0
00)
(-0.0
99)
(-0.1
51)
(-0.1
08)
Schooling(F
-H)
0.2
93***
0.4
18***
0.2
53***
0.1
54
0.3
66**
0.1
17
(-0.0
13)
(-0.0
16)
(-0.0
16)
(-0.1
42)
(-0.1
29)
(-0.1
56)
Inflate
Dgeo
H,F
1.2
41***
0.5
96*
1.7
15***
(-0.1
70)
(-0.2
79)
(-0.2
79)
Dcult
H,F
4.5
29***
5.7
75***
3.9
03***
(-0.2
89)
(-0.4
37)
(-0.5
14)
Population
H-0
.000***
-0.0
00***
-0.0
00***
(0.0
00)
(0.0
00)
(0.0
00)
Population
F-0
.000***
-0.0
00***
-0.0
00***
(0.0
00)
(0.0
00)
(0.0
00)
Schooling(F
-H)
0.2
24***
0.6
55***
0.1
71***
(-0.0
27)
(-0.0
56)
(-0.0
48)
N218526
198660
19866
51667
46970
4697
r2p
--
-0.3
18
0.3
27
0.4
57
Est
imationmethod:
Zero
Inflate
dB
inom
ial
Regre
ssio
ns
inm
odels
(1)
to(3
);P
ois
son
pse
udo-m
axim
um
likelihood
est
imato
r(P
PM
L)
inm
odels
(4)
to(6
);
Dependentvariable:
tota
lnum
ber
of
M&
Adeals
from
Hto
Fin
models
(1)
to(3
)to
tal
valu
eof
M&
Adeals
from
Hto
Fin
models
(1)
to(3
)
All
regre
ssors
inm
odels
(4)
to(6
)are
inlo
gs.
t-st
ati
stic
sin
pare
nth
ese
s.*p<
0.0
5,
**p<
0.0
1,
***p<
0.0
01.
10
C Linguistic and religious distance: comparison with available mea-
sures
As explained in Section 3, our bilateral measure of cultural distance is an index obtained as a weighted average
of three distance measures: genetic, linguistic, and religious distances. While genetic distance is from Spolaore
and Wacziarg (2009), linguistic and religious distances are our own computation based on information drawn from
Ethnologue and the CIA World Factbook, respectively.
The linguistic distance matrix is fully derived by applying Eq. (1) to information drawn from the phylogenetic
linguistic tree provided by ‘Ethnologue: Languages of the World.’ The phylogenetic tree is a diagram reflecting the
tree model of language origination. The first level of the tree consists of a certain number of language families.4 A
family is a monophyletic unit in which all members derive from a common ancestor; all attested descendants of that
ancestor are included in the family. Language families can be divided into smaller phylogenetic units (‘branches’).
The position of each language in the tree is identified by a code from which a common number of branches can
be identified. The maximum number of branches in the Ethnologue tree is 15. As an example, since English and
Standard German share tree branches (3.5.2.1.1 is the code for English and 3.5.2.3.1.1.1.1 is the code for Standard
German), their distance, according to Eq. (1), amounts to 1−(
315
)0.5.
The international distribution of languages is obtained from Ethnologue.
The final index of linguistic distance is calculated based on 6855 languages distributed across 195 countries. The
index is available, together with the other measures of distance, on http://docenti.unich.it/delgatto/delgatto_
web/research.htm. A replication package with all the data and STATA codes can also be downloaded.
Recently, two linguistic distance measures, based on Fearon’s procedure and using the Ethnologue data, which are
thereby comparable with ours, have been made available for download. The first is used in Melitz and Toubal (2014) and is
distributed through the CEPII website (variable LP1). The second is described by Spolaore and Wacziarg (2016) and can
be downloaded from the authors’ website.
For the three measures, Table A3 reports the summary statistics obtained after averaging the values by country (for
each country, we consider the average bilateral distance to the other countries).
Spolaore and Wacziarg (2016) have information on 156 countries and for 7 of them, the linguistic distance is always 1
(the maximum value). Compared to our measure, the distribution looks more skewed toward high values.
As for the measure developed by Melitz and Toubal (2014), a key difference with respect to ours is that their measure
subsumes the differences between languages in four possible cases: 0 for two languages belonging to separate family trees,
0.25 for two languages belonging to different branches of the same family tree (English and French), 0.50 for two languages
belonging to the same branch (English and German), and 0.75 for two languages belonging to the same sub-branch (German
and Dutch). Consequently, the resulting index is less articulated than ours: in 35 of the 195 countries under consideration
(the index takes a value of zero whenever two languages belong to different families), the index of linguistic proximity takes
a value of zero (i.e., the maximum value) and the overall variability is lower than ours.
Our index covers 195 countries and there are no countries for which bilateral distances are always maximum or minimum.
A comprehensive list of the countries covered by the three datasets is reported in Table 20.
4Ethnologue identifies 141 different language families (i.e., top-level genetic groups). Six of these (namely, Afro-Asiatic, Austronesian,Indo-European, Niger-Congo, Sino-Tibetan, and Trans-New Guinea), each of which has at least 5% of the speakers of the world’slanguages, stand out as the major language families of the world. Together, they account for nearly two-thirds of all languages andfive-sixths of the world’s population.
11
As recognized by Spolaore and Wacziarg, “a drawback of tree-based measures is that linguistic distance is calculated on
a discrete number of common nodes, which could be an imperfect measure of separation times between languages. A single
split between two languages that occurred a long time ago would result in the same measure of distance than a more recent
single split, but the languages in the first case may in fact be more distant than in the second. Similarly, numerous recent
splits may result in two languages sharing few nodes, while a smaller number of very distant linguistic subdivisions could
make distant languages seem close.” (Spolaore and Wacziarg, 2015, p. 12) To overcome these limitations, other measures
have been developed. Spolaore and Wacziarg (2016) describe in detail measures based on the answers provided to the World
Values Survey. Melitz and Toubal (2014) rely on a linguistic proximity indicator drawn from the Automated Similarity
Judgment Program, which provides an index of similarities of words with identical meanings for a limited vocabulary of
words between different language pairs based on expert judgments. However, the number of countries for which these
measures are available is in general much lower, which is why we rely only on tree-based measures.
For religious distance, we use the same tree as Spolaore and Wacziarg (2009), which Roman Wacziarg and James Fearon
kindly provided us, while the international distribution of religions was drawn from the CIA World Factbook.
Table A4 highlights that, although the distributions of our index of religious distance look quite similar to the one in
Spolaore and Wacziarg (2016), our measure includes more countries and features a higher variability, probably because of
the different data on the international distribution of religions. A comprehensive list of the countries covered by the two
matrixes is reported in Table 21.
12
Tab
leA
2:C
ult
ura
ld
ista
nce
:to
pfo
ur
cult
ura
lly
sim
ilar
cou
ntr
ies
by
acqu
irin
gco
untr
y.
Countr
yA
Countr
yB
Cult
ura
ldis
tance
1if
conti
guous
Countr
yA
Countr
yB
Cult
ura
ldis
tance
1if
conti
guous
Aust
ralia
Unit
ed
Kin
gdom
0.0
171
1It
aly
Fra
nce
0.2
128
1A
ust
ralia
New
Zeala
nd
0.0
349
1It
aly
Bra
zil
0.2
247
0A
ust
ralia
Irela
nd-R
ep
0.0
473
1Japan
Thailand
0.5
689
0A
ust
ralia
Unit
ed
Sta
tes
0.0
969
1Japan
Chin
a0.5
865
0A
ust
ria
Germ
any
0.1
129
1Japan
South
Kore
a0.6
016
0A
ust
ria
Sw
itzerl
and
0.2
445
1Japan
Sin
gap
ore
0.6
192
0A
ust
ria
Neth
erl
ands
0.2
686
1N
eth
erl
ands
Aust
ria
0.2
686
1A
ust
ria
Irela
nd-R
ep
0.2
730
0N
eth
erl
ands
Germ
any
0.2
767
1B
elg
ium
Neth
erl
ands
0.2
086
1N
eth
erl
ands
Aust
ralia
0.2
869
0B
elg
ium
Fra
nce
0.2
458
1N
eth
erl
ands
Irela
nd-R
ep
0.2
920
0B
elg
ium
Spain
0.2
850
0N
ew
Zeala
nd
Aust
ralia
0.0
349
1B
elg
ium
Port
ugal
0.2
850
0N
ew
Zeala
nd
Unit
ed
Kin
gdom
0.0
395
1C
anada
Aust
ralia
0.1
146
1N
ew
Zeala
nd
Unit
ed
Sta
tes
0.1
179
1C
anada
Unit
ed
Kin
gdom
0.1
227
1N
ew
Zeala
nd
Canada
0.1
390
0C
anada
Irela
nd-R
ep
0.1
367
1N
orw
ay
Denm
ark
0.1
841
1C
anada
New
Zeala
nd
0.1
390
1N
orw
ay
Sw
eden
0.1
917
1D
enm
ark
Norw
ay
0.1
841
1N
orw
ay
Germ
any
0.3
219
1D
enm
ark
Sw
eden
0.1
930
1N
orw
ay
Unit
ed
Kin
gdom
0.3
237
0D
enm
ark
Germ
any
0.3
211
1P
ort
ugal
Bra
zil
0.0
025
1D
enm
ark
Unit
ed
Kin
gdom
0.3
239
0P
ort
ugal
Spain
0.1
198
1F
inla
nd
Est
onia
0.3
750
1P
ort
ugal
Pola
nd
0.3
443
0F
inla
nd
Hungary
0.3
943
1P
ort
ugal
Germ
any
0.3
740
0F
inla
nd
Sw
eden
0.5
007
1Spain
Uru
guay
0.0
408
1F
inla
nd
Norw
ay
0.5
245
1Spain
Arg
enti
na
0.0
633
1Fra
nce
Port
ugal
0.1
721
1Spain
Cuba
0.0
846
1Fra
nce
Spain
0.1
754
1Spain
Colo
mbia
0.0
945
0Fra
nce
Uru
guay
0.1
800
1Sw
eden
Norw
ay
0.1
917
1Fra
nce
Bra
zil
0.1
956
0Sw
eden
Denm
ark
0.1
930
1G
erm
any
Aust
ria
0.1
129
1Sw
eden
Icela
nd
0.2
540
1G
erm
any
Sw
itzerl
and
0.2
591
1Sw
eden
Germ
any
0.3
274
1G
erm
any
Neth
erl
ands
0.2
767
1Sw
itzerl
and
Aust
ria
0.2
445
1G
erm
any
Aust
ralia
0.2
850
0Sw
itzerl
and
Germ
any
0.2
591
1G
reece
Ukra
ine
0.4
030
1Sw
itzerl
and
Fra
nce
0.2
707
1G
reece
Rom
ania
0.4
059
0Sw
itzerl
and
Neth
erl
ands
0.3
059
0G
reece
Bulg
ari
a0.4
170
0U
nit
ed
Kin
gdom
Aust
ralia
0.0
171
1G
reece
Macedonia
0.4
172
0U
nit
ed
Kin
gdom
New
Zeala
nd
0.0
395
1Ir
ela
nd-R
ep
Aust
ralia
0.0
473
1U
nit
ed
Kin
gdom
Irela
nd-R
ep
0.0
629
1Ir
ela
nd-R
ep
Unit
ed
Kin
gdom
0.0
629
1U
nit
ed
Kin
gdom
Unit
ed
Sta
tes
0.1
011
1Ir
ela
nd-R
ep
Unit
ed
Sta
tes
0.1
316
1U
nit
ed
Sta
tes
Aust
ralia
0.0
969
1Ir
ela
nd-R
ep
Canada
0.1
367
0U
nit
ed
Sta
tes
Unit
ed
Kin
gdom
0.1
011
1It
aly
Port
ugal
0.2
000
1U
nit
ed
Sta
tes
New
Zeala
nd
0.1
179
1It
aly
Spain
0.2
113
1U
nit
ed
Sta
tes
Irela
nd-R
ep
0.1
316
1
13
Table A3: Index of linguistic distance: comparison of Spolaore and Wacziarg (2016) and Melitz and Toubal (2014)
Melitz and Toubal (2014)Percentiles Smallest
1% 0 05% 0 010% 0 0 Obs 19525% 0.1 0 Sum of Wgt. 195
50% 0.8 Mean 0.6487432Largest Std. Dev. 0.5816996
75% 1.1 1.66500790% 1.4 1.815939 Variance 0.338374495% 1.5 1.855484 Skewness 0.146718499% 1.9 1.882626 Kurtosis 1.463548
Spolaore and Wacziarg (2016)Percentiles Smallest
1% 0.9 0.88205655% 0.9 0.887934310% 0.9 0.8926941 Obs 15625% 1 0.9024695 Sum of Wgt. 156
50% 1 Mean 0.9719247Largest Std. Dev. 0.0244978
75% 1 190% 1 1 Variance 0.000600195% 1 1 Skewness -1.53478199% 1 1 Kurtosis 5.308946
Our indexPercentiles Smallest
1% 0.8 0.82522535% 0.8 0.825225310% 0.9 0.8266248 Obs 19525% 0.9 0.8274217 Sum of Wgt. 195
50% 0.9 Mean 0.9251468Largest Std. Dev. 0.0449256
75% 1 0.996498590% 1 0.9969382 Variance 0.002018395% 1 0.9969657 Skewness -0.464850799% 1 0.9999439 Kurtosis 2.59972
14
Table A4: Index of religious distance: comparison with Spolaore and Wacziarg (2016).
Spolaore and Wacziarg (2016)Percentiles Smallest
1% 0.7491348 0.74201485% 0.7654374 0.749134810% 0.7817982 0.7519308 Obs 15625% 0.818274 0.7585266 Sum of Wgt. 156
50% 0.8598031 Mean 0.8584846Largest Std. Dev. 0.0545077
75% 0.9002472 0.964641790% 0.9278184 0.9731171 Variance 0.002971195% 0.9501613 0.9769342 Skewness -0.02293199% 0.9769342 0.9812025 Kurtosis 2.408189
Our IndexPercentiles Smallest
1% 0.6265798 0.62657985% 0.6265798 0.626579810% 0.6279002 0.6265798 Obs 19525% 0.6348516 0.6265798 Sum of Wgt. 195
50% 0.707397 Mean 0.7115864Largest Std. Dev. 0.0895338
75% 0.7450541 0.965309890% 0.8290228 0.9660383 Variance 0.008016395% 0.9426194 0.9660496 Skewness 1.32070999% 0.9660496 0.9701512 Kurtosis 4.247559
15
Min Max Min Max Min MaxAfghanistan 0.601132 1 0 3.891733 0.9365966 1Albania 0.6307231 1 0 1.945866 0.8589453 1Algeria 0.4789545 1 0 5.837599 0.8164966 1Angola 0.2748575 1 0 5.837599 1 1Antigua and Barbuda 0.1223147 1 0 3.891733Argentina 0.0337567 1 0 5.837599 0.0995034 1Armenia 0.7511756 1 0 1.945866 0.9578312 1Australia 0.0178127 1 0 3.891733 0.1103723 1Austria 0.2315434 1 0 5.837599 0.3090368 1Azerbaijan 0.6485394 1 0 3.269056 0.9148143 1Bahamas 0.2948545 1 0 3.891733Bahrain 0.3258855 1 0 5.837599 0.8049845 1Bangladesh 0.5560532 1 0 3.891733 0.8828081 1Barbados 0.4516945 1 0 3.891733Belarus 0.5617793 1 0 4.086319 0.8487618 1Belize 0.5781972 1 0 3.34689Benin 0.5551797 1 0 0 0.8489221 1Bermuda 0.2948545 1 0 3.891733Bhutan 0.7329378 1 0 1.75128 0.8659644 1Brazil 0.025367 1 0 5.837599 0 1Bulgaria 0.5999311 1 0 3.891733 0.9046388 1Burkina Faso 0.5306003 1 0 0 0.8735732 1Burundi 0.1843011 1 0 3.891733 0.5773503 1Cameroon 0.4181035 1 0 0 0.8239278 1Canada 0.2000898 1 0 3.307973 0.5383005 1Cape Verde 0.6585576 1 0 3.891733Central African Republic 0.6160944 1 0 0 0.8815502 1Chad 0.867963 1 0 0 0.9220702 1Chile 0.0153058 1 0 5.837599 0 1China 0.5589733 1 0 1.070227 0.4794149 1Colombia 0.0180865 1 0 5.837599 0 1Comoros 0.272656 1 0 5.837599Costa Rica 0.023918 1 0 5.837599 0.1428572 1Croatia 0.4913954 1 0 5.837599 0.9081706 1Cuba 0 1 0 5.837599 0.0999999 1Cyprus 0.2465824 1 0 1.537235 0.4584368 1Czech Republic 0.5422176 1 0 5.837599 0.9436892 1Denmark 0.430337 1 0 5.837599 0.7745967 1Djibouti 0.2853348 1 0 0 0.9499445 1Dominica 0.1984033 1 0 5.837599Dominican Republic 0.0233717 1 0 5.837599 0 1Ecuador 0.1418713 1 0 5.837599 0 1Egypt 0.4492703 1 0 5.837599 0.1417761 1El Salvador 0.0082664 1 0 5.837599 0 1Eritrea 0.644192 1 0 3.891733 1 1Estonia 0.658197 1 0 2.840965 0.8365045 1Fiji 0.6001015 1 0 1.945866 0.9555105 1Finland 0.658197 1 0 2.840965 0.9414484 1France 0.3796932 1 0 5.837599 0.3249468 1Gabon 0.2275005 1 0 0 0.740755 1Georgia 0.9343106 1 0 0 0.9677145 1Germany 0.2315434 1 0 5.837599Ghana 0.4885217 1 0 0 0.850944 1Greece 0.2465824 1 0 1.945866 0.9520463 1Greenland 0.8591734 1 0 5.837599Grenada 0.4369094 1 0 3.891733Guatemala 0.4879393 1 0 5.837599 0.663325 1Guinea 0.5577977 1 0 0 0.8727937 1Guyana 0.6893518 1 0 3.891733 0.6875896 1Haiti 0.624027 1 0 5.837599 0.7745967 1Honduras 0.0264587 1 0 5.837599 0.1466471 1Hungary 0.7457266 1 0 1.945866 0.9473763 1Iceland 0.5537792 1 0 5.837599India 0.6172843 1 0 0 0.9248534 1Indonesia 0.6365855 1 0 3.891733 0.8970595 1Iraq 0.5331682 1 0 5.837599 0.4748498 1Ireland 0.0650404 1 0 3.891733 0.9349844 1Israel 0.5637879 1 0 0.5312214 0.7777162 1Italy 0.2907522 1 0 3.891733 0.8563488 1Jamaica 0.5540789 1 0 3.891733 0.3465336 1Japan 0.9945585 1 0 0 1 1Jordan 0.4378405 1 0 5.837599 0.2394368 1Kazakhstan 0.5479229 1 0 2.514254 0.7844148 1Kenya 0.4902678 1 0 5.837599 0.7960392 1Kiribati 0.3225637 1 0 3.891733Kuwait 0.3046775 1 0 5.837599 0.7949461 1Kyrgyzstan 0.6662565 1 0 2.840965 0.8788032 1Latvia 0.5539984 1 0 3.385808 0.8141656 1Lebanon 0.2558222 1 0 5.837599 0.7876299 1Liberia 0.6529831 1 0 0 0.9319282 1Lithuania 0.5539984 1 0 3.385808 0.9258971 1Madagascar 0.6357519 1 0 1.945866 0.8961299 1Malawi 0.2761755 1 0 3.891733 0.6531972 1Malaysia 0.6659513 1 0 2.607461 0.8394758 1Mali 0.6131939 1 0 0 0.8690562 1Malta 0.5879993 1 0 5.837599Mauritania 0.4693248 1 0 5.837599 0.8484744 1Mauritius 0.7701765 1 0 5.837599 0.9192018 1
Our index Melitz-Toubal (2014) Spolaore-Wacziarg (2015)
Mexico 0.0680444 1 0 5.837599 0.1252673 1Morocco 0.5138625 1 0 5.837599 0.8523507 1Mozambique 0.2554138 1 0 0 0.6892869 1Nepal 0.6266349 1 0 3.891733 0.9136314 1Netherlands 0.4284745 1 0 5.837599 0.8988882 1Netherlands Antilles 0.7235321 1 0 5.837599New Caledonia 0.5817834 1 0 5.837599New Zealand 0.0235442 1 0 3.891733 0.4691953 1Nicaragua 0.0429515 1 0 5.837599 0.2236068 1Niger 0.8370821 1 0 0 0.9695034 1Nigeria 0.7253903 1 0 0 0.9208692 1Norway 0.430337 1 0 5.837599 0.8215674 1Oman 0.3793254 1 0 5.837599 0.4950654 1Pakistan 0.5758918 1 0 3.891733 0.9279807 1Panama 0.1856813 1 0 5.837599 0.1466471 1Papua New Guinea 0.8751732 1 0 0Paraguay 0.9532863 1 0 0 1 1Peru 0.2211512 1 0 5.837599 0 1Philippines 0.6508641 1 0 1.945866 0.9814199 1Poland 0.562347 1 0 3.891733 0.8973128 1Portugal 0.025367 1 0 5.837599 0.68313 1Qatar 0.4228716 1 0 5.837599Romania 0.2070801 1 0 3.891733 0.9096477 1Rwanda 0.1843011 1 0 3.891733 0.68313 1Sao Tome and Principe 0.6585576 1 0 5.837599Saudi Arabia 0.4225538 1 0 5.837599 0.4219803 1Senegal 0.5513123 1 0 1.945866 0.9301646 1Seychelles 0.6473948 1 0 5.837599Sierra Leone 0.6371534 1 0 0 0.9529988 1Singapore 0.6043872 1 0 1.023526 0.9213296 1Slovakia 0.5422176 1 0 5.837599 0.8741308 1Slovenia 0.4913954 1 0 5.837599 0.8262433 1Solomon Islands 0.4379734 1 0 0Somalia 0.2853348 1 0 1.945866 0.7148305 1South Africa 0.3479718 1 0 0 0.7473043 1Spain 0.1197343 1 0 5.837599 0.4397123 1Sri Lanka 0.585974 1 0 3.50256 0.9507262 1Sudan 0.6008883 1 0 5.837599 0.8525953 1Suriname 0.6395724 1 0 4.903584Sweden 0.4486489 1 0 5.837599 0.9327321 1Switzerland 0.5080295 1 0 4.825749 0.8011405 1Taiwan 0.5589733 1 0 1.070227 1 1Thailand 0.57307 1 0 5.837599 0.7916228 1Togo 0.4885217 1 0 0 0.8926896 1Tonga 0.1837693 1 0 1.945866Trinidad and Tobago 0.6199217 1 0 3.891733 0.6753542 1Tunisia 0.4242706 1 0 5.837599 0.7860477 1Turkey 0.6226946 1 0 3.269056 0.9250069 1Turkmenistan 0.6096752 1 0 3.269056 0.9068432 1Uganda 0.5096226 1 0 0 0.8178677 1Ukraine 0.5617793 1 0 4.086319 0.7947614 1United Arab Emirates 0.3046775 1 0 5.837599 0.9420407 1Uruguay 0 1 0 5.837599 0 1Uzbekistan 0.6096752 1 0 3.269056 0.9380861 1Vanuatu 0.3693715 1 0 0Zambia 0.2767366 1 0 0 0.6788306 1Zimbabwe 0.2386447 1 0 3.891733 0.6947421 1Cambodia 0.6487034 1 0 3.891733Yemen 0.424408 1 0 5.837599American Samoa 0.022531 1Belgium 0.4284745 1 0.8529574 1Bolivia 0.5789325 1 0.7491492 1Botswana 0.2330396 1 0.6193295 1Brunei 0.6365855 1Cote d'Ivoire 0.5146606 1 0.894869 1Equatorial Guinea 0.2275005 1Ethiopia 0.6276844 1 0.9171333 1French Guiana 0.6937249 1French Polynesia 0.3446193 1Guadeloupe 0.0536571 1Guam 0.5212091 1Guinea Bissau 0.621531 1 0.8462576 1Iran 0.6529881 1 0.9029343 1Laos 0.57307 1 1 1Lesotho 0.2289814 1 0.6202783 1Libya 0.7282829 1 0.7880133 1Luxembourg 0.3743481 1Macedonia 0.6342529 1Maldives 0.5560532 1Martinique 0.0536571 1Moldova 0.2070801 1 0.655131 1Mongolia 0.7407994 1 0.9671745 1Namibia 0.4101656 1 0.7985259 1Puerto Rico 0.0275477 1Reunion 0.6364548 1San Marino 0.2907522 1Swaziland 0.2289814 1 0.6567597 1Syria 0.2558222 1 0.3578582 1Tanzania 0.3062449 1 0.6994997 1United Kingdom 0.0178127 1 0.5126991 1Venezuela 0.013318 1 0.9660918 1
Vietnam 0.6487034 1 0.9903958 1Yugoslavia 0.6307231 1 0.4848392 1Congo 0 0 0.7324676 1Gibraltar 0 3.463642Russian Federation 0 4.086319 0.6647279 1Tajikistan 0 3.891733 0.8657795 1Congo, Dem. Rep. 0.43622 1Congo, Rep. 0.4829018 1Gambia, The 0.5513123 1Korea, North 0 1Korea, South 0 1Myanmar 0.7830979 1Russia 0.5479229 1Samoa 0.022531 1St. Kitts and Nevis 0.1223147 1St. Lucia 0.1984033 1St. Vincent and the Grenadines 0.4272408 1Tajikstan 0.601132 1United States 0.1025335 1Virgin Islands 0.5036657 1Andorra 0 5.020335Anguilla 0 3.891733Aruba 0 5.837599Belgium and Luxembourg 0 4.164155Bolivia (Plurinational State of) 0 3.152304Bosnia and Herzegovina 0 5.837599British Virgin Islands 0 3.891733Brunei Darussalam 0 3.891733Cayman Islands 0 3.891733China, Hong Kong Special Administrative Region 0 1.070227Cook Islands 0 0Côte d'Ivoire 0 0Democratic Republic of the Congo 0 0Falkland Islands (Malvinas) 0 3.891733Gambia 0 0Guinea-Bissau 0 0Iran (Islamic Republic of) 0 2.802048Lao People's Democratic Republic 0 5.837599Libyan Arab Jamahiriya 0 5.837599Marshall Islands 0 3.891733Micronesia (Federated States of) 0 0Montserrat 0 3.891733Nauru 0 3.891733Niue 0 1.945866Norfolk Island 0 3.891733Northern Mariana Islands 0 0Palau 0 1.945866Pitcairn 0 3.891733Republic of Korea 0 0Republic of Moldova 0 3.891733Saint Helena 0 3.891733Saint Kitts and Nevis 0 3.891733Saint Lucia 0 5.837599Saint Pierre and Miquelon 0 3.891733Saint Vincent and the Grenadines 0 3.891733Syrian Arab Republic 0 5.837599The former Yugoslav Republic of Macedonia 0 3.891733Turks and Caicos Islands 0 3.891733Tuvalu 0 1.945866United Kingdom of Great Britain and Northern Ireland 0 3.891733United Republic of Tanzania 0 5.837599United States of America 0 3.599853Venezuela (Bolivarian Republic of) 0 5.837599Viet Nam 0 3.891733Channel IslandsCzechoslovakia 0.5602881 1Faeroe IslandsGerman Democratic Republic 0.9660918 1Germany, Federal Republic of 0.6086931 1Hong KongIsle of ManKampuchea, Democratic 0.9076005 1Korea 1 1Korea,Dem.Rep. 0 1MacaoMyanmar(Burma) 0.9672489 1St Christopher and NevisSt LuciaSt. VincentThe Gambia 0.8480354 1U.S.A 0.8993456 1U.S.S.R. 0.9448028 1Virgin Islands (U.S.)Western SamoaYemen, Arab Republic of 0.9403723 1Yemen, People's Democratic Republic ofZaire 0.714349 1
Min Max Min MaxCambodia 0.6487034 1Congo, Dem. Rep. 0.43622 1Congo, Rep. 0.4829018 1Gambia, The 0.5513123 1Korea, North 0 1Korea, South 0 1Myanmar 0.7830979 1Russia 0.5479229 1Samoa 0.022531 1St. Kitts and Nevis 0.1223147 1St. Lucia 0.1984033 1St. Vincent and the Grenadines 0.4272408 1Tajikstan 0.601132 1United States 0.1025335 1Virgin Islands 0.5036657 1Yemen 0.424408 1Channel IslandsCongo 0.7068239 1Czechoslovakia 0.7085337 0.9977975Faeroe IslandsGerman Democratic Republic 0.8041393 1Germany, Federal Republic of 0.6523802 1GibraltarHong KongIsle of ManKampuchea, Democratic 0.296648 1Korea 0.7332121 0.9787952Korea,Dem.Rep. 0.7138347 0.9992497MacaoMyanmar(Burma) 0.4117039 0.9934989Russian Federation 0.747556 0.9976974St Christopher and NevisSt LuciaSt. VincentTajikistan 0.4862098 1The Gambia 0.2863564 1U.S.A 0.6239872 0.9919677U.S.S.R.Virgin Islands (U.S.)Western SamoaYemen, Arab Republic of 0.6253799 0.9941328Yemen, People's Democratic Republic of 0.8455767 0.9899495Zaire 0.6403124 0.9904544Afghanistan 0.601132 1 0.3162278 1Albania 0.6307231 1 0.5196153 1Algeria 0.4789545 1 0.1000002 1American Samoa 0.022531 1Angola 0.2748575 1 0.7834029 1Antigua and Barbuda 0.1223147 1Argentina 0.0337567 1 0.2863565 1Armenia 0.7511756 1 0.6659729 1Australia 0.0178127 1 0.654981 1Austria 0.2315434 1 0.4523273 1Azerbaijan 0.6485394 1 0.2449489 1Bahamas 0.2948545 1Bahrain 0.3258855 1 0.4341659 1Bangladesh 0.5560532 1 0.3464101 0.9988994Barbados 0.4516945 1Belarus 0.5617793 1 0.4613025 1Belgium 0.4284745 1 0.3316625 1Belize 0.5781972 1Benin 0.5551797 1 0.6752778 1Bermuda 0.2948545 1Bhutan 0.7329378 1 0.4816638 1Bolivia 0.5789325 1 0.2029778 1Botswana 0.2330396 1 0.7071068 1Brazil 0.025367 1 0.4664761 0.9983987Brunei 0.6365855 1Bulgaria 0.5999311 1 0.6889993 0.9994999Burkina Faso 0.5306003 1 0.6928203 1Burundi 0.1843011 1 0.5859522 1Cameroon 0.4181035 1 0.7523297 1Canada 0.2000898 1 0.5837808 0.9983587Cape Verde 0.6585576 1Central African Republic 0.6160944 1 0.7549834 0.9951382Chad 0.867963 1 0.6708204 1
Our index Spolaore-Wacziarg (2015)
Chile 0.0153058 1 0.2534561 1China 0.5589733 1 0.7339755 0.9925523Colombia 0.0180865 1 0.3461214 0.9981984Comoros 0.272656 1Costa Rica 0.023918 1 0.2924721 0.998018Cote d'Ivoire 0.5146606 1 0.5966573 1Croatia 0.4913954 1 0.7933851 0.8388087Cuba 0 1 0.410731 0.9982985Cyprus 0.2465824 1 0.4433509 0.9982585Czech Republic 0.5422176 1 0.9153797 0.9229301Denmark 0.430337 1 0.3466698 0.998018Djibouti 0.2853348 1 0.6437391 1Dominica 0.1984033 1Dominican Republic 0.0233717 1 0.2662705 0.9980982Ecuador 0.1418713 1 0.2662705 0.9980982Egypt 0.4492703 1 0.2190891 1El Salvador 0.0082664 1 0.3986477 0.9982785Equatorial Guinea 0.2275005 1Eritrea 0.644192 1 0.7519308 0.7519308Estonia 0.658197 1 0.7178301 1Ethiopia 0.6276844 1 0.6557438 0.9966946Fiji 0.6001015 1 0.8014362 0.9979379Finland 0.658197 1 0.4586502 0.997998France 0.3796932 1 0.3223041 1French Guiana 0.6937249 1French Polynesia 0.3446193 1Gabon 0.2275005 1 0.6689544 0.9973966Georgia 0.9343106 1 0.6975958 0.9982284Germany 0.2315434 1Ghana 0.4885217 1 0.7989994 0.9939618Greece 0.2465824 1 0.6464055 0.998028Greenland 0.8591734 1Grenada 0.4369094 1Guadeloupe 0.0536571 1Guam 0.5212091 1Guatemala 0.4879393 1 0.3550211 0.9980381Guinea 0.5577977 1 0.3660601 1Guinea Bissau 0.621531 1 0.7345747 1Guyana 0.6893518 1 0.748131 0.9967447Haiti 0.624027 1 0.3500856 0.9980782Honduras 0.0264587 1 0.2265831 0.9980581Hungary 0.7457266 1 0.4323886 0.9981382Iceland 0.5537792 1India 0.6172843 1 0.5166043 0.9981583Indonesia 0.6365855 1 0.3224903 0.9988193Iran 0.6529881 1 0.5133225 0.9990095Iraq 0.5331682 1 0.6218038 0.9990295Ireland 0.0650404 1 0.2973212 0.9981182Israel 0.5637879 1 0.8282512 0.9989495Italy 0.2907522 1 0.2219009 1Jamaica 0.5540789 1 0.7134844 0.9986991Japan 0.9945585 1 0.428859 0.9991496Jordan 0.4378405 1 0.2280352 1Kazakhstan 0.5479229 1 0.6618157 0.998609Kenya 0.4902678 1 0.6870226 0.997998Kiribati 0.3225637 1Kuwait 0.3046775 1 0.6244998 0.9942334Kyrgyzstan 0.6662565 1 0.4582576 0.9937304Laos 0.57307 1 0.9528903 1Latvia 0.5539984 1 0.7435321 0.9984989Lebanon 0.2558222 1 0.7278736 1Lesotho 0.2289814 1 0.5385165 1Liberia 0.6529831 1 0.7523297 1Libya 0.7282829 1 0.1732049 0.9948367Lithuania 0.5539984 1 0.5546891 1Luxembourg 0.3743481 1Macedonia 0.6342529 1Madagascar 0.6357519 1 0.7302055 1Malawi 0.2761755 1 0.6395311 1Malaysia 0.6659513 1 0.697137 0.9720494Maldives 0.5560532 1Mali 0.6131939 1 0.3130494 1Malta 0.5879993 1Martinique 0.0536571 1Mauritania 0.4693248 1 0.0894427 1Mauritius 0.7701765 1 0.720708 0.9764425Mexico 0.0680444 1 0.3133051 0.9966444Moldova 0.2070801 1 0.6832276 1
Mongolia 0.7407994 1 0.2939388 0.9983186Morocco 0.5138625 1 0.1604992 1Mozambique 0.2554138 1 0.745654 1Namibia 0.4101656 1 0.6136774 1Nepal 0.6266349 1 0.8366122 0.9984388Netherlands 0.4284745 1 0.7124605 0.9969353Netherlands Antilles 0.7235321 1New Caledonia 0.5817834 1New Zealand 0.0235442 1 0.7190132 1Nicaragua 0.0429515 1 0.4132796 0.9958413Niger 0.8370821 1 0.4608687 0.9939819Nigeria 0.7253903 1 0.6517668 1Norway 0.430337 1 0.4809989 0.997998Oman 0.3793254 1 0.8117881 0.9937304Pakistan 0.5758918 1 0.3872983 0.9948367Panama 0.1856813 1 0.2905168 1Papua New Guinea 0.8751732 1 0.7144228 1Paraguay 0.9532863 1 0.2561252 1Peru 0.2211512 1 0.3509986 0.9958916Philippines 0.6508641 1 0.3594441 0.9951583Poland 0.562347 1 0.276767 1Portugal 0.025367 1 0.2917534 0.9959317Puerto Rico 0.0275477 1Qatar 0.4228716 1Reunion 0.6364548 1Romania 0.2070801 1 0.7946823 1Rwanda 0.1843011 1 0.554617 0.9955803San Marino 0.2907522 1Sao Tome and Principe 0.6585576 1Saudi Arabia 0.4225538 1 0.228035 0.997998Senegal 0.5513123 1 0.3040394 1Seychelles 0.6473948 1Sierra Leone 0.6371534 1 0.6246599 1Singapore 0.6043872 1 0.5105488 0.9914434Slovakia 0.5422176 1 0.6581489 0.88Slovenia 0.4913954 1 0.6232175 0.8820431Solomon Islands 0.4379734 1Somalia 0.2853348 1 0.2683282 1South Africa 0.3479718 1 0.6910861 0.9987993Spain 0.1197343 1 0.2582248 0.997998Sri Lanka 0.585974 1 0.5505089 0.9903737Sudan 0.6008883 1 0.5498182 1Suriname 0.6395724 1Swaziland 0.2289814 1 0.7099296 1Sweden 0.4486489 1 0.6748333 0.9958614Switzerland 0.5080295 1 0.5598929 0.9963031Syria 0.2558222 1 0.4788319 1Taiwan 0.5589733 1 0.6587867 0.9947361Tanzania 0.3062449 1 0.7451174 1Thailand 0.57307 1 0.340441 0.9981683Togo 0.4885217 1 0.7395945 1Tonga 0.1837693 1Trinidad and Tobago 0.6199217 1 0.7764277 0.9699278Tunisia 0.4242706 1 0.2540867 0.997998Turkey 0.6226946 1 0.4356145 1Turkmenistan 0.6096752 1 0.365568 0.9959919Uganda 0.5096226 1 0.6717142 1Ukraine 0.5617793 1 0.798724 0.9955401United Arab Emirates 0.3046775 1 0.6936858 0.9947864United Kingdom 0.0178127 1 0.677761 0.9919677Uruguay 0 1 0.4442521 0.99Uzbekistan 0.6096752 1 0.3783649 0.9943842Vanuatu 0.3693715 1Venezuela 0.013318 1 0.4519735 0.9959919Vietnam 0.6487034 1 0.5905759 0.9904141Yugoslavia 0.6307231 1Zambia 0.2767366 1 0.819451 0.980714Zimbabwe 0.2386447 1 0.8146778 0.997998