Professional Documents
Culture Documents
Article PDI September 2014
Article PDI September 2014
Article PDI September 2014
Abstract
I. Introduction
China would thus topple the United States, which held the limelight as the
world’s largest economy since overtaking the United Kingdom in 1872.
However, some economists have expressed skepticism regarding China’s
long-term prospects (Fan et al, 2012). China is suffering from
overinvestment and high leverage, the Time Magazine noted in April 2014.
More importantly, “the U.S. still has a substantial lead in innovation, and its
dominant position in many industries and sectors is not about to vanish.”2
Briefly, results of this study are as follows. First, it shows that there is a
two-way interaction between cultural characteristics and economic
outcomes. Second, impact of cultural propensity to build hierarchical
relationships affects economic performance through two channels – ability
to innovate and competence in building effective institutions. Quality of
1
World Bank press release. April 29, 2014
2
The Time magazine. April 30, 2014
institutions and innovation should therefore be viewed as endogenous to
cultural characteristics. Third, countries with high power distance index can
compensate for tendency to build hierarchical relationships by openness
and oil rents to ahieve economic prosperity. Finally, findings of this study
cast doubt on China’s ability to take place of the United States of America
as the next global centre of capitalism. The latter finding rationalizes
China’s links with global diasporas in developed countries and focus on
acquisition of technologies in developed markets.
T
Power Distance Index=α 1 +β 1 ×Log (GDP) per capita+∑1 β T × Χ T +e 1
N
Log (GDP ) per capita=α2 + β2 ×Power Distance Index + ∑1 β N ×Γ N +e 2 (1),
where Χ T and Γ N are vectors that contain control variables and errors are correlated .
Sample used in this study includes 93 countries, for which both power
distance index and confucian dynamism (confucian LTO) are available.
Variables used in this study and sources of data are described in Appendix
A. Logarithm of GDP per capita in 2012 is used to compare attained wealth.
One characteristic problem for studies that examine economic growth and
wealth is causality and endogeneity (Paldam, 1996). As Barro (1996, 1999),
this study uses an instrumental-variable technique, where some of the
instruments are earlier values of regressors.
Global Innovation Index was first estimated in 2007, but coverage for 2008
is more comprehensive, so innovation index is measured as of 2008. As a
robustness check, government efficiendy index for 2012 and innovation
index for 2013, the latest dates available at the time of completion of this
study, are tested in robustness checks.
III. Results
Univariates
Wealth regressions
Reverse causality
The question of impact of culture on economic outcomes does not meet with
a uniform response in academic literature. Alexander and Seidman (1990)
are critical of the concept of a unique national culture, whereas McSweeney
(2002) is skeptical about Hofstede’s research methods and, more generally,
about the plausibity of relationship between national culture and uniform
national institutions.
This study does not assume that national culture is deterministic, but it
assumes that culture, political and economic life are intertwined and have
mutual effects. There is a two-way interaction between culture and
economic outcomes. Linton (1936) speaks of culture as “the precipitate of
history,” suggesting that historic events have impact on national psyche.
Culture could thus affect economic outcomes.
Potter (1954) argues that the United States of America’s ideal of freedom
and democracy is largely based on accumulation of wealth that
accompanied democratic choices. Freedom and abundance, in Potter’s
words, “are to such a great extent fused in American democratic thought”
(Potter, 1954: 127). Further, “democracy is clearly most appropriate for
countries which enjoy an economic surplus and least appropriate for
countries where there is an economic insufficiency. In short, economic
abundance is conducive to political democracy” (Potter, 1954: 127).
Power distance index attains larger significance than in models that do not
incorporate a two-way interaction between between cultural and economic
variables in table 2. In addition to standard choice of variables for this
paper, model 5 – the first stage model that explains logarithm of GDP per
capita – incorporates percentage of spending on research and development,
but PDI retains its negative sign and significance in the presence of
innovation index.
Oil rents have positive impact on hierarchical culture, and large countries
tend to have more hierarchies in place. As expected, hierarchies are
stronger in less educated and less wealthy societies.
First stage regression models 7-9 in table 2 employ univariate models with
power distance index as the only independent variable to estimate
innovation and government efficiency. It it possible that PDI attains
significance because it proxies for omitted variables in these models.
Oil rents take on a positive sign at low level of government efficiency in line
with Rodriguez and Sachs (1999), and Sachs and Warner (2001), who show
that countries with greater natural resources wealth tend to have a lower
quality of public sector governance.
Results from table 5 highlight difficulties that lie ahead of China in its
developmental drive. They imply that trend extrapolation for China’s growth
rates should not be implemented – China is facing bottlenecks in the form of
inefficient institutions and hierarchical culture that stifles innovation.
Table 6 reproduces the regressions of table 2 for high and low hierarchy
country subsamples. Models 1-5 are for high PDI subsample, and models 6-9
are for low PDI subsample to allow direct comparison. The focus is not on
the PDI variable, but on variables that measure level of education and
foreign trade impact. Coefficient on schooling variable is slightly higher for
high power distance index countries, but most of the differences can be
found in variables that measure foreign trade impact. Coefficients on both
trade openness and oil rents are not significant in low PDI subsample, but
they attain high significance in countries with hierarchical culture.
Apparently, hierarchies may be tempered by an openness to ideas and an
open economy, as in China. Alternatively, countries with hierarchical
cultures could use natural resource endowment to get to higher levels of
wealth.
Robustness
Law and finance literature (La Porta et al, 1997, 1998) has recently
achieved academic prominence. One key finding of this literature is that
“law matters”, in the sense that various aspects of national legal systems
shape economic outcomes. This finding was later questioned on premises
that the observed correlation does not necessarily imply a causal
relationship (Rajan and Zingales, 1988) and that, contrary to what the law
and finance literature would predict, in the 1930s-1970s the U.S. stock
market development was not triggered by improvements in investor
protection, but by extralegal factors (Cheffins et al, 2012).
Without taking a side in this debate, this study attempts to disentangle legal
environment effects by separating the sample of 93 countries into four
categories by legal origin: English (28), French (43), German (15), and
Scandinavian (5). Models in table 7 take on the same specification as
models reported in table 2. Control variables suppressed. Beta coefficients
on PDI variable and, in models 6-9, beta coefficients on innovation index
and government efficiency index in second/third stage models are reported.
PDI variable takes on the same signs in all subsamples, except a subsample
of Scandinavian countries, which includes just five observations.
Results from table 7 validate major findings of this paper. Further, they
suggest that results are not based on outliers. Low statistical power of legal
origin variables in the presence of cultural variables in table 7 suggests
another avenue of research in the area of “law and finance”. Cultural
preferences could explain why certain countries design elaborate legal
system and at the same time develop highly efficient financial institutions
and markets based on “external finance”. If two processes occur
independently driven by cultural preferences, correlation between legal
choices and economic outcomes could be spurious. This would explain why
development of legal framework in the U.S. in the 1930s was not followed
by higher stock market valuations (Cheffins et al, 2012).
Several ways to calibrate power distance index have been attempted in this
paper to ensure robustness of results. One set of approach uses indicator
variables set to one if power distance index exceeds a certain threshold. A
second approach uses PDI as a continuous variable. More complicated
specifications have been implemented, including quadratic and piecewise
linear terms. These add no significant explanatory power. The relationship
between PDI and our dependent variables is monotonic.
IV. Conclusion
These findings cast a serious doubt on the premise that historically high
growth rates in China could extrapolate into the future. They formalize the
argument that China’s growth would be slowed down by weakness of its
institutions and low innovation levels.
References
Fan, J., R. Morck, and B. Yeung. 2012. Capitalizing China. NBER Working
Paper No. 17687
Sachs, J., and A. Warner. 1995. Economic Reform and the Process of Global
Integration. Brookings Papers on Economic Activity, I, 1-118
Sachs, J., and A. Warner. 2001. The Curse of Natural Resources, European
Economic Review, 45, 827-838
Schuman, M. 2014. China Could Overtake the U.S. as the World’s No. 1
Economy This Year. Time Magazine, April 30, 2014
Smith, A. 1776. An Inquiry into the Nature and Causes of the Wealth of
Nations. London, UK: Methuen, 1904
Weber, M. 1905. The Protestant Ethic and the Spirit of Capitalism, New
York: Scribner's Press, 1958
Wind, J., Mahajan, V., and J. L. Bayless. 1990. The Role of New Product
Models in Supporting and Improving the New Product Development
Process: Some Preliminary Results. Cambridge, MA: Marketing Science
Institute
Power distance index (PDI) Geert and Gert Jan Hofstede website.
Coefficient is scaled by 100
Panel B. Frequencies for high PDI and low PDI subsamples Chi-
Square
N N
statistic
High Income Dummy 17 2 for
14.92***
Colony 23 29 frequenc
3.87**
ies
Tabletable
This 1. Selected
reports data for low
selected income group
characteristics and high
of high powerincome group
distance countries
index and low power
distance index countries. Each subsample includes 46 bservations. Power distance index
threshold of 0.675 separates high/low PDI subsamples. Panel A reports mean and median Z-
for each subsample, t-statistic for means and Z-statistic for medians. Panel B statistic
reports
T-
frequency distributions and Chi-square stiatistic.∗, ∗∗, and ∗∗∗ indicate a p-value of for
10%,
Panel
5%, and
A. 1%,
Meanrespectively.
and median for high PDI and low PDI subsamples statistic
differenc
for
e
differenc
Mean Median Mean Median in
e
medians
in
Power Distance Index 0.46 0.48 0.78 0.80 -12.25***-9.54***
means
Confucian (LTO) 0.45 0.37 0.40 0.37 0.32 0.00
GDP per capita 2012 24,213 18,771 8,968 3,899 4.47*** 4.05***
Log (GDP per capita) 2012 9.60 9.84 8.23 8.27 4.99*** 3.64***
Log (GDP per capita) 1980 9.04 9.51 7.51 7.51 4.74*** 3.06***
Innovation Index 2008 43.3 42.5 35.1 34.3 4.55*** 3.90***
Innovation Index 2013 46.7 49.4 35.9 36.1 4.89*** 4.31***
Government Effectiveness Index
75.3 81.5 46.4 50.0 5.68*** 4.75***
2000
Government Effectiveness Index
75.2 83.0 47.8 47.0 5.51*** 4.85***
2012
Corruption 2000 73.4 81.0 44.3 46.0 5.57*** 4.81***
Freedom of Press 2000 11.67 6.5 33.12 25.5 -4.89*** -4.72***
Rule of Law 2000 72.1 79.2 43.3 42.6 5.52*** 4.87***
R&D expense (% of GDP) average 19 1.13 0.78 0.34 0.03 4.66*** 4.68***
Education (% of GDP) average 1970- 4.5 4.5 3.8 3.6 2.17*** 2.08**
Total Schooling 2000 9.1 9.4 7.1 7.0 4.19*** 3.74***
Log (Fertility) 2000 3.1 3.0 3.5 3.5 1.96* 2.09**
Log (Life Expectancy) 2000 4.3 4.3 4.2 4.3 2.95*** 3.66***
Trade Openness 1991-2000 71.2 63.7 79.9 60.0 0.82 0.36
Oil rents 1991-2000 1.3 0.0 7.6 1.0 3.36*** 2.85***
Log (Population, mln) 2000 2.7 2.4 3.1 3.1 1.33 1.02
Panel B. Frequencies for high PDI and low PDI subsamples Chi-
Square
N N
statistic
High Income Dummy 17 2 for
14.92***
Colony 23 29 frequenc
3.87**
ies
Tabletable
This 1. Selected
reports data for low
selected income group
characteristics and high
of high powerincome group
distance countries
index and low power
distance index countries. Each subsample includes 46 bservations. Power distance index
threshold of 0.675 separates high/low PDI subsamples. Panel A reports mean and median
for each subsample, t-statistic for means and Z-statistic for medians. Panel B reports
T-
frequency distributions and Chi-square stiatistic.∗, ∗∗, and ∗∗∗ indicate a p-value of 10%,
Panel
5%, and
A. 1%,
Mean respectively.
and median for high PDI and low PDI subsamples statistic
for
differenc
Mean Median Mean Median
e
in
Power Distance Index 0.46 0.48 0.78 0.80 -12.25***
means
Confucian (LTO) 0.45 0.37 0.40 0.37 0.32
GDP per capita 2012 24,213 18,771 8,968 3,899 4.47***
Log (GDP per capita) 2012 9.60 9.84 8.23 8.27 4.99***
Log (GDP per capita) 1980 9.04 9.51 7.51 7.51 4.74***
Innovation Index 2008 43.3 42.5 35.1 34.3 4.55***
Innovation Index 2013 46.7 49.4 35.9 36.1 4.89***
Government Effectiveness Index
75.3 81.5 46.4 50.0 5.68***
2000
Government Effectiveness Index
75.2 83.0 47.8 47.0 5.51***
2012
Corruption 2000 73.4 81.0 44.3 46.0 5.57***
Freedom of Press 2000 11.67 6.5 33.12 25.5 -4.89***
Rule of Law 2000 72.1 79.2 43.3 42.6 5.52***
R&D expense (% of GDP) average 19 1.13 0.78 0.34 0.03 4.66***
Education (% of GDP) average 1970- 4.5 4.5 3.8 3.6 2.17***
Total Schooling 2000 9.1 9.4 7.1 7.0 4.19***
Log (Fertility) 2000 3.1 3.0 3.5 3.5 1.96*
Log (Life Expectancy) 2000 4.3 4.3 4.2 4.3 2.95***
Trade Openness 1991-2000 71.2 63.7 79.9 60.0 0.82
Oil rents 1991-2000 1.3 0.0 7.6 1.0 3.36***
Log (Population, mln) 2000 2.7 2.4 3.1 3.1 1.33
Panel B. Frequencies for high PDI and low PDI subsamples Chi-
Square
N N
statistic
for
frequenc
ies
Tabletable
This 1. Selected
reports data for low
selected income group
characteristics and high
of high powerincome group
distance countries
index and low power
distance index countries. Each subsample includes 46 bservations. Power distance index
threshold of 0.675 separates high/low PDI subsamples. Panel A reports mean and median Z-
for each subsample, t-statistic for means and Z-statistic for medians. Panel B statisticreports
T-
frequency distributions and Chi-square stiatistic.∗, ∗∗, and ∗∗∗ indicate a p-value of for
10%,
Panel
5%, and
A. 1%,
Meanrespectively.
and median for high PDI and low PDI subsamples statistic
differenc
for
e
differenc
Mean Median Mean Median in
e
medians
in
Power Distance Index 0.46 0.48 0.78 0.80 -12.25***-9.54***
means
Confucian (LTO) 0.45 0.37 0.40 0.37 0.32 0.00
GDP per capita 2012 24,213 18,771 8,968 3,899 4.47*** 4.05***
Log (GDP per capita) 2012 9.60 9.84 8.23 8.27 4.99*** 3.64***
Log (GDP per capita) 1980 9.04 9.51 7.51 7.51 4.74*** 3.06***
Innovation Index 2008 43.3 42.5 35.1 34.3 4.55*** 3.90***
Innovation Index 2013 46.7 49.4 35.9 36.1 4.89*** 4.31***
Government Effectiveness Index
75.3 81.5 46.4 50.0 5.68*** 4.75***
2000
Government Effectiveness Index
75.2 83.0 47.8 47.0 5.51*** 4.85***
2012
Corruption 2000 73.4 81.0 44.3 46.0 5.57*** 4.81***
Freedom of Press 2000 11.67 6.5 33.12 25.5 -4.89*** -4.72***
Rule of Law 2000 72.1 79.2 43.3 42.6 5.52*** 4.87***
R&D expense (% of GDP) average 19 1.13 0.78 0.34 0.03 4.66*** 4.68***
Education (% of GDP) average 1970- 4.5 4.5 3.8 3.6 2.17*** 2.08**
Total Schooling 2000 9.1 9.4 7.1 7.0 4.19*** 3.74***
Log (Fertility) 2000 3.1 3.0 3.5 3.5 1.96* 2.09**
Log (Life Expectancy) 2000 4.3 4.3 4.2 4.3 2.95*** 3.66***
Trade Openness 1991-2000 71.2 63.7 79.9 60.0 0.82 0.36
Oil rents 1991-2000 1.3 0.0 7.6 1.0 3.36*** 2.85***
Log (Population, mln) 2000 2.7 2.4 3.1 3.1 1.33 1.02
Panel B. Frequencies for high PDI and low PDI subsamples Chi-
Square
N N
statistic
High Income Dummy 17 2 for
14.92***
Colony 23 29 frequenc
3.87**
ies
Table 1. Selected
This table reports data for low
selected income group
characteristics and high
of high powerincome group
distance countries
index and low power
distance index countries. Each subsample includes 46 bservations. Power distance index
threshold of 0.675 separates high/low PDI subsamples. Panel A reports mean and median
for each subsample, t-statistic for means and Z-statistic for medians. Panel B reports
T-
frequency distributions and Chi-square stiatistic.∗, ∗∗, and ∗∗∗ indicate a p-value of 10%,
Panel
5%, and
A. 1%,
Meanrespectively.
and median for high PDI and low PDI subsamples statistic
for
differenc
Mean Median Mean Median
e
in
Power Distance Index 0.46 0.48 0.78 0.80 -12.25***
means
Confucian (LTO) 0.45 0.37 0.40 0.37 0.32
GDP per capita 2012 24,213 18,771 8,968 3,899 4.47***
Log (GDP per capita) 2012 9.60 9.84 8.23 8.27 4.99***
Log (GDP per capita) 1980 9.04 9.51 7.51 7.51 4.74***
Innovation Index 2008 43.3 42.5 35.1 34.3 4.55***
Innovation Index 2013 46.7 49.4 35.9 36.1 4.89***
Government Effectiveness Index
75.3 81.5 46.4 50.0 5.68***
2000
Government Effectiveness Index
75.2 83.0 47.8 47.0 5.51***
2012
Corruption 2000 73.4 81.0 44.3 46.0 5.57***
Freedom of Press 2000 11.67 6.5 33.12 25.5 -4.89***
Rule of Law 2000 72.1 79.2 43.3 42.6 5.52***
R&D expense (% of GDP) average 19 1.13 0.78 0.34 0.03 4.66***
Education (% of GDP) average 1970- 4.5 4.5 3.8 3.6 2.17***
Total Schooling 2000 9.1 9.4 7.1 7.0 4.19***
Log (Fertility) 2000 3.1 3.0 3.5 3.5 1.96*
Log (Life Expectancy) 2000 4.3 4.3 4.2 4.3 2.95***
Trade Openness 1991-2000 71.2 63.7 79.9 60.0 0.82
Oil rents 1991-2000 1.3 0.0 7.6 1.0 3.36***
Log (Population, mln) 2000 2.7 2.4 3.1 3.1 1.33
Panel B. Frequencies for high PDI and low PDI subsamples Chi-
Square
N N
statistic
High Income Dummy 17 2 for
14.92***
Colony 23 29 frequenc
3.87**
ies
Table 1. Selected
This table reports data for low
selected income group
characteristics and high
of high powerincome group
distance countries
index and low power
distance index countries. Each subsample includes 46 bservations. Power distance index
threshold of 0.675 separates high/low PDI subsamples. Panel A reports mean and median
for each subsample, t-statistic for means and Z-statistic for medians. Panel B reports
T-
frequency distributions and Chi-square stiatistic.∗, ∗∗, and ∗∗∗ indicate a p-value of 10%,
Panel
5%, A. 1%,
and Mean and median for high PDI and low PDI subsamples
respectively. statistic
for
differenc
Mean Median Mean Median
e
in
Power Distance Index 0.46 0.48 0.78 0.80 -12.25***
means
Confucian (LTO) 0.45 0.37 0.40 0.37 0.32
GDP per capita 2012 24,213 18,771 8,968 3,899 4.47***
Log (GDP per capita) 2012 9.60 9.84 8.23 8.27 4.99***
Log (GDP per capita) 1980 9.04 9.51 7.51 7.51 4.74***
Innovation Index 2008 43.3 42.5 35.1 34.3 4.55***
Innovation Index 2013 46.7 49.4 35.9 36.1 4.89***
Government Effectiveness Index
75.3 81.5 46.4 50.0 5.68***
2000
Government Effectiveness Index
75.2 83.0 47.8 47.0 5.51***
2012
Corruption 2000 73.4 81.0 44.3 46.0 5.57***
Freedom of Press 2000 11.67 6.5 33.12 25.5 -4.89***