Okui Causality Between Economic and Political Freedom
Transcript of Okui Causality Between Economic and Political Freedom
-
8/7/2019 Okui Causality Between Economic and Political Freedom
1/38
1
Causality between Political Freedom
and Economic Freedom
March 7, 2005
Katsuyoshi Okui
Abstract
The aim of this paper is to investigate causality between political freedom and economicfreedom. Since economic freedom data do not have long time series, I conduct
causality tests that use panel data with only short time-series but many cross-sectionalcountries.
A glance at the data shows that political freedom promotes economic
freedom. Granger causality tests that do not take account of individual effects leadto the results that political freedom causes economic freedom. But my causality
tests that take account of individual effects lead to the results that there is nocausality between political freedom and economic freedom.
Otemon Gaku in University, 2-1-15, Nishiai, Ibarak i, Osaka , 567-8502, J apan E-ma il: [email protected]
-
8/7/2019 Okui Causality Between Economic and Political Freedom
2/38
2
1. Introduction
The aim of this paper is to investigate causality between political freedom and
economic freedom. Here political freedom means how degree democratization has
progressed. Economic freedom means how degree marketization has progressed.
Does democratization cause marketization? Or, does marketization cause
democratization? I want to investigate them by estimating political-economic freedom
autoregressions and conducting causality tests.
When we think about development strategy of developing countries, we often
face the question which we should put priority on between democratization and
marketization. Results of this paper are considered to contribute to solving this
question. For example, if we get the result that democratization causes marketization
and marketization does not cause democratization, we get the implication that
democratization should be put priority on.
The paper is organized as follows. Section 2 presents a background of the
issue that I deal with, and refers to some prior researches. Section 3 reviews previous
works that pursue the relationship between political freedom and economic freedom.
Section 4 explains political and economic freedom data that I use and overviews the
data. It is shown that that there are more number of countries where marketization
progressed after democratization had progressed than the ones where democratization
progressed after marketization had progressed. To check it using econometric method,
I conduct Granger causality tests between political freedom and economic freedom in
section 5. In section 6, weak points of the Granger causality tests are pointed out and
another type of causality tests is explained. I conduct this type of causality tests and
show their results in section 7. In section 8, I conduct causality tests using different
data of political freedom and economic freedom. Section 9 discusses the conclusion
and policy implication.
2. BackgroundThe collapse of Soviet East Europe socialistic countries in 1980's, Asian
currency crisis in 1997, and default problems of emerging countries and/or the like have
made the following old questions be new and important. What kind of political and
economic state system is desirable? In order to head for this system, what sort of
process should be taken?
Around the questions, many people agree little resistance that developing
democratization and progressing marketization are preferable. But when it comes to
strategy of developing democratization and progressing marketization, opinions about
-
8/7/2019 Okui Causality Between Economic and Political Freedom
3/38
3
the strategy are largely divided. One of main differences of the opinions is the one
about which should be put priority on between marketization or democratization.
Some say that marketization should be accelerated, thinking highly of efficie ncy. Theothers say that developing democratization and enlarging freedom of choice should be
more urgent than realizing economic growth. People who believe market force, who
advocate big-ban approach about transition to market economy such as Sachs, and who
support current IMF policy, development economists such as Lal who emphasize
government failure, and the like can be said to be close to the former. People who
consider market failure is important and who look much to government role, Sen who
thinks much of democracy instead of economic growth and insists that entitlement
should be given to the poor, and the like can be said to be close to the latter.
This difference is considered to start disappearing if causality between
marketization and democratization is known. For example, if democratization tend to
promote marketization and marketization does not tend to promote democratization, we
can say that the policy that put more priority on democratization than on marketization
is preferable. The aim of this paper is to deepen understanding causality between
political freedom and economic freedom and to get a clue to construct desirable regime.
As seen in "economic freedom is also indispensable means toward the
achievement of political freedom (Friedman, 1962, 8)," the relationship between
political freedom and economic freedom has been aware of long before. But it seems
that there have been few works that make theoretical models dealing with the
relationship between political freedom and economic freedom. Hence, taking
advantage of Granger-causality technique seems to be a good idea because the
technique enables us to explore the causality without theoretical models.
But there is one problem left. Since economic freedom data have just begun
to be made recently, it is difficult to collect enough time series data to conduct
Granger-causality tests. Economic freedom data of the Fraser Institute has the longest
time series. The data are reported every five years from 1970 to 2000, although thedata have been reported every year since 2000. Thus period that the data cover is so
short that we cannot conduct usual time-series version of Granger-causality tests.
However, the data are made for the purpose of country-by-country comparison and have
a relatively large cross-section part. By taking advantage of the characteristic of the
panel data that have a short time-series part and a large cross-section part, I want to
conduct causality tests. Holtz-Eakin, Newey, and Rosen (1988, 1989) propose a useful
method to conduct a causality test with panel data. They say that it is not necessary to
collect long time series data when applying their method into causality tests. Hence
-
8/7/2019 Okui Causality Between Economic and Political Freedom
4/38
4
we conduct causality tests of political freedom and economic freedom by owing much
to their method.
3. Literature Review
Recently there have been many works that explore the relationship between
politico-economic regime and economic performance (for example, Scully and Slottje
(1991), De Haan and Sierman (1995, 1998), Barro (1996), Vanssay and Spindler (1994),
Knack and Keefer (1995), Eaton and Walker (1997), Durham (1999), Chong and
Calderon (2000), Ali and Crain (2001) and so on.) That is, they examine the
followings. Does political freedom affects economic growth and per capita GDP?
How about economic freedom? However, there have not been so many works that
explore the relationship between political freedom and economic freedom.
Wu and Otto (1999) investigate the relationship among the four (political
freedom, economic freedom, economic growth, and economic development) by using a
technique named log-linear model. They say that clear correlation between political
freedom and economic freedom is not found. Dethier et al. (1999) and De Haan and
Sturm (2003) investigate whether political freedom facilitates economic freedom for
former communist countries and developing countries respectively. Both conclude
that political freedom facilitates economic freedom. Dawson (1998) investigates the
relationship between economic performance and institutions by using standard
regression analysis. Dawson also checks the relationship between political freedom
and economic freedom in the paper and get the conclusion that the initial level of
political freedom affects economic freedom and that change in economic freedom
affects political freedom. Dawson (2003) investigates the similar relationship by using
Granger causality tests. The conclusion is that political freedom causes economic
freedom and that economic freedom does not cause economic freedom, although both
directions of causality are observed in the case of summary indexes of both freedom and
some other cases.Feng (2003) directly addresses the relationship between political freedom and
economic freedom and conducts Granger causality tests. Feng's result is that political
freedom causes economic freedom and that economic freedom does not cause political
freedom. But Feng does not take country-specific factors into account. It may occur
bias and hide the true relationship because the country-specific factors can make
significantly different levels of political freedom and economic freedom.
Farr, Lord, and Wolfenbarger (1998) investigate causality among political
freedom, economic freedom, and per capita GDP by conducting Granger causality tests.
-
8/7/2019 Okui Causality Between Economic and Political Freedom
5/38
5
In an effort to deal with individual effects, country dummy variables are included in
their estimated regressions. Their conclusion is that there is no causality between the
two. They use economic freedom data of the old version of the Fraser Institute.They estimate vector autoregressions assuming that the lag length is one. And it is not
very clear how degree individual effects work in their results. The method proposed
by Holtz-Eakin, Newey and Rosen does not require us to assume lag length and enables
us to check whether we can neglect individual effects or not. I conduct causality tests
by using the new version of the Fraser Institute and the method by Holtz-Eakin, Newey
and Rosen.
4. Political Freedom Data and Economic Freedom Data
I would like to begin by explaining political freedom data and economic
freedom data that we use for our analysis.
4.1. Explanation about both freedom data
Political freedom data shows how degree a country progresses democratization. I use
political rights index and civil liberties index issued by Freedom House as political
freedom data. Political rights index shows how degree fair and meaningful elections
are executed. Civil liberties index implies freedom of press, freedom of speech,
freedom of religious belief, and the right to protest and organize. Higher values of
both indexes mean less democratic. Freedom House has issued these indexes annually
since 1973. Because there are seven years' economic freedom data such as 1970, 1975,
1980, 1985, 1990, 1995, and 2000, I prepare political freedom data of 1975, 1980, 1985,
1990, 1995, and 2000. In order to make each year's political freedom data, we average
five years before and after the year's data instead of using the year's data itself. For
example, when we make 1975 political right index, we average 1973, 1974, 1975, 1976,
1977 political right indexes. This is because evaluation of political institution seems to
be better by making use of information before and after the year's data. Now we getpolitical rights index and civil liberties index of 1975, 1980, 1985, 1990, 1995, and
2000. Then I add up these two indexes and multiply it by minus 1 so that freer
countries have higher value indexes. Next I standardize them so that their average
becomes zero and their standard deviation becomes 1. I use this political freedom
index as political freedom data.
Economic freedom data shows how degree a country progresses marketization.
I use economic freedom data issued by the Fraser Institute. The data see as
economically free the situation where personal choice, protection of private property,
-
8/7/2019 Okui Causality Between Economic and Political Freedom
6/38
6
and freedom of exchange are ensured. Gwartney, Lawson et al. (2002) say that their
Chain-Weighted Summary index has the highest quality among some economic freedom
data. I use the Chain-Weighted Summary index and standardize it. I use the index aseconomic freedom data.
4.2. Overviewing both freedom data
Let us overview the data made in the above process. As explained before, I use the
political freedom index and the economic freedom index of 1975, 1980, 1985, 1990,
1995, and 2000 for our analysis. Because there are at best 123 countries that have
economic freedom data, I choose countries for the political freedom index taking
account of it. Basic statistics of the political freedom index are shown in Table 1.
Basic statistics of the economic freedom index are shown in Table 2.
Table 1- Basic statistics of the political freedom index
Year 1975 1980 1985 1990 1995 2000 AllN 113 114 114 121 123 123 708
M ean -0.25 -0.16 -0.12 0.04 0.18 0.27 0.00M ax 1.37 1.37 1.37 1.37 1.37 1.37 1.37M in -1.76 -1.76 -1.76 -1.71 -1.76 -1.76 -1.76SD 1.05 1.02 1.05 0.98 0.91 0.90 1.00
Table 2- Basic statistics of the economic freedom index
Year 1975 1980 1985 1990 1995 2000 AllN 79 111 114 119 123 123 669
M ean -0.32 -0.33 -0.29 -0.17 0.29 0.65 0.00M ax 2.20 2.28 2.02 1.95 2.52 2.36 2.52M in -2.24 -2.24 -2.74 -2.20 -1.61 -1.65 -2.74SD 1.01 0.95 0.98 0.96 0.91 0.80 1.00
Movement of averages of the political freedom index is shown in Figure 1. From it,we can tell development of democratization of entire world. And movement of
averages of the economic freedom index is shown in Figure 2. We can see that
marketization has been progressed rapidly since around 1990.
-
8/7/2019 Okui Causality Between Economic and Political Freedom
7/38
7
Figure 1. Movement of the political freedom index Figure 2. Movement of the economic freedom index
What country has developed democratization well? What country has a high economic
freedom? Let us show the best and worst 10 countries of both 2000 indexes. Table 3
and Table 4 describe them. Japan is 22nd among 123 countries for political freedom
and 27th among 123 countries for economic freedom.
Table 3. High-democratic countries and low-democratic countries
Dem ocraticcountries
politicalfreedomindex
Non-dem ocraticcountries
politicalfreedomindex
Australia 1.37 Syria -1.76Austria 1.37 M yanm ar -1.76Barbados 1.37 C hina -1.54Belize 1.37 Rwanda -1.48C anada 1.37 C ongo,Dem .R. -1.43Denm ark 1.37 Burundi -1.39Finland 1.37 Bahrain -1.38Iceland 1.37 C am eroon -1.32Ireland 1.37 Iran -1.30Luxem bourg 1.37 Egypt -1.18M alta 1.37Netherlands 1.37New Zealand 1.37Norway 1.37Portugal 1.37Sweden 1.37Sw itzerland 1.37United States 1.37
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
1975 1980 1985 1990 1995 2000
Political Freedom IndexEconom ic Freedom Index
-0.40
-0.20
0.00
0.20
0.40
0.60
0.80
1975 1980 1985 1990 1995 2000
-
8/7/2019 Okui Causality Between Economic and Political Freedom
8/38
8
Table 4. High- economic-free countries and low economic-free countries
Econom ic-fee
countries
econom ic
freedomindex
Non-
econom icfree
countries
econom ic
freedomindex
Hong Kong 2.36 M yanm ar -1.65Singapore 2.22 C ongo,Dem .R. -1.60United States 2.17 G uinea-B issau -1.21United Kingdo 2.03 Algeria -1.09Sw itzerland 1.95 Ukraine -0.75New Zealand 1.92 Russia -0.57Ireland 1.87 C ongo,Rep.of -0.54
C anada 1.80 M alawi -0.51Netherlands 1.79 Zim babw e -0.50Australia 1.78 Togo -0.49
4.3. The relationship between political freedom and economic freedom
Table 5 shows the relationship between movement of political freedom and movement
of economic freedom. In how many countries did the political freedom index or the
economic freedom index increase or decrease during each four period of 1975-1980,
1980-1985, 1985-1990, and 1990-1995? Among the countries where it increases or
decreases, in how many countries did the other freedom index increase or decrease
during the next 5-year-period? For example, there are 113 countries where political
freedom data exist in 1975 and 1980. Among the 113 countries, there are 56 countries
where political freedom index increased during 1975-1980. Among these 56 countries,
there are 32 countries where economic freedom index increased during 1980-1985.
Since the ratio is 21 countries among 31 countries, it is 32/56=0.57. Table 6 is the
same as Table 5 with the exception of criteria that freedom index increase or decrease
by 0.1. For example, there are 43 countries where political freedom index increasedby 0.1 during 1975-1980. Among these 43 countries, there are 21 countries where
economic freedom index increased by 0.1 during 1980-1985. Since the ratio is 21
countries among 43 countries, it is 21/43=0.49.
What is immediately obvious from the table is that the ratio of the countries
where economic freedom increased after democratization had been progressed is higher
than ones of the other cases. It seems to suggest that democratization promotes
marketization. There seems to be causality from political freedom to economic
freedom. Further below, we check causality between political freedom and economic
-
8/7/2019 Okui Causality Between Economic and Political Freedom
9/38
9
freedom more rigorously by using econometric methods. In below analysis, I turn 79
countries into an object of our analysis since the 79 countries have complete data of
both freedom indexes for six years.
Table 5. The relationship between movement of political freedom and movement economic freedom
(criteria is just increase or decrease)
19751980 19801985 19851990 19901995 All
A number of countries113 114 114 121 462
The number of countries where democratization was progressed
56 43 61 52 212Among the above countries, the number of countries where
economic freedom increased during the next period32 32 50 45 159
Ratio 0.57 0.74 0.82 0.87 0.75
A number of countries113 114 114 121 462
The number of countries where democratization was progressed 29 37 30 54 150
Among the above countries, the number of countries where
economic freedom increased during the next period10 13 5 14 42
Ratio 0.34 0.35 0.17 0.26 0.28
A number of countries79 111 114 119 423
The number of countries where democratization was progressed 50 65 77 97 289
Among the above countries, the number of countries where
economic freedom increased during the next period19 37 31 49 136
Ratio 0.38 0.57 0.40 0.51 0.47
A number of countries79 111 114 119 423
The number of countries where democratization was progressed 28 42 35 20 125
Among the above countries, the number of countries whereeconomic freedom increased during the next period 14 13 15 8 50
Ratio 0.50 0.31 0.43 0.40 0.40
-
8/7/2019 Okui Causality Between Economic and Political Freedom
10/38
-
8/7/2019 Okui Causality Between Economic and Political Freedom
11/38
11
Each a and d is a parameter, m is a lag length, and ut is a disturbance. I conduct
panel type of Granger causality tests. Granger causality tests are conducted
through the following procedure. If the null hypothesis x does not cause y isrejected,x Granger-causesy. That is, if the null hypothesis: 1 =2 ==m=0
is rejected, we judge thatx affectsy. We can apply a standard F test into testing the
hypothesis. I conduct the F test taking the freedom variables of 1995 and 2000 as
explained variables. And I check the cases where the lag lengths are three, two,
and one, i.e., m=3, 2, 1.
Results about the hypothesis economic freedom does not cause political
freedom are shown in Table 7. In the case of m=3, I regress political freedom on
three lagged political freedom and three lagged economic freedom using ordinary
least squares technique. Obtained sum of squared residuals is 22.45. Degrees of
freedom in this case is 151 because the number of data 158 (79 observations for two
period) minus the number of parameters 7. Next I estimate the equation in which
three lagged economic freedom are excluded from the above equation. Thus
obtained sum of squared residuals is 23.55. The Fstatistics from these values is
.
This Fvalue has an F-distribution with 3 degrees of freedom of numerator and with
151 degrees of freedom of denominator under the hypothesis: 1=2=3=0 .
The value 2.47 is the one of which the upper side of the F distribution is 6.4 percent,
shown as p. Hence we cannot reject the null hypothesis at the 1-percent level and
5-percent level but at 10-percent level. Although we can slightly see causality from
economic freedom to political freedom, the causality is not so strong. The cases of
m=1 and m=2 show similar results. It is difficult to say that economic freedom
causes political freedom.
Table 7. Results of Granger-causality tests:
economic freedom political freedom
Sum of squared
residualsF
Degrees of
Freedom
m =3 22.45 -----
ExcludeE 23.55 2.47 3,151 0.064 *
m =2 22.94 -----
ExcludeE 23.73 2.65 2,153 0.074 *
m =1 23.46 -----
ExcludeE 24.00 3.52 1,155 0.063 *
***significant at the 1 -percent level
** Significant at the 5-percent level* Significant at the 10-percent level
F =
23.55-22.45 /3
22.45/151= 2.47
-
8/7/2019 Okui Causality Between Economic and Political Freedom
12/38
12
Results about the hypothesis political freedom does not cause economic freedom are
shown in Table 8. The null hypothesis is strongly rejected in all cases ofm=3, 2, 1.
Political freedom causes economic freedom. These results of the simple F testsshow that political freedom affects economic freedom more largely than economic
freedom affects political freedom.
Table 8. Results of Granger-causality tests:
political freedom economic freedom
Sum of squared
residualsF
Degrees of
Freedom
m =3 27.02 -----
Exclude P 29.27 4.19 3,151 0.007 ***
m =2 27.16 -----
Exclude P 29.36 6.22 2,153 0.003 ***
m =1 27.18 -----
Exclude P 29.37 12.52 1,155 0.001 ***
***significant at the 1 -percent level** significant at the 5-percent level* significant at the 10-percent level
6. Another Type of Causality Tests Proposed by Holtz-Eakin=Newey=Rosen
Now we get the result that political freedom causes economic freedom, while it
is difficult to say that economic freedom causes political freedom. The result is
consistent with the fact in section 4 that there are more countries where economic
freedom increased after democratization had been progressed. And the result
supports the recent works that investigate causality between political freedom and
economic freedom, such as the Dawson (2003) and Feng (2003). Does political
freedom really cause economic freedom? Does political freedom really have more
effect on economic freedom than economic freedom does on political freedom? I will
look more carefully to these points.One of the biggest problems about the tests that I conducted above is to neglect
country-specific factors named individual effects. The tests assume that every country
has the same intercept. It means that the individual effects are not taken into account.
The individual effect summarizes the influence of unobserved variables which have
persistent effect on the dependent variables. Therefore the intercept and the dependent
variable are correlated if there is the individual effect. The correlation result in
inconsistent estimates. Consider the following equation that adds an individual effect
(fi) into equation the (1).
-
8/7/2019 Okui Causality Between Economic and Political Freedom
13/38
13
yit = a o +
m
l=1
a lt yit-l +
m
l=1
dlt xtit-l + fi + uit
(2 )A standard method of eliminating the individual effect is to first difference the data to
eliminate fi and then use ordinary or generalized squares to estimate the differenced
equation:
yit yit-1 =
m
l=1
al yt-l yt-l-1 +
m
l=1
dl xtt-l xtt-l-1 + (uit uit-1 )
(3 )
This approach, however, induce a simultaneity problem. Becauseyit-1depends on uit-1,
(uit uit-1) is correlated with the regressor (yit-1 yit-2). A solution of this problem is to
employ instrumental variables that are uncorrelated with (uit uit-1) such asyit-2, yit-3, ,yi1,
x it-2, xit-3, , xi1. Holtz-Eakin, Newey, and Rosen (1988, 1989) propose estimation
technique that allows the existence of an individual effect and that employs the
instrumental variables. I will use the similar method to theirs.
Consider the following model:
yit = aot +
m
l=1
a lt yit-l +
m
l=1
dlt xtit-l + fi + uitt=1,...,T
i=1,...,N
(4)
where fi is an unobserved individual effect. Subtracting the t-1 period from the t
period of equation (4), we get the following equation.
D yit = at +
m+1
l=1
clt yit-l +
m+1
l=1
dlt xtit-l + vit t= m+2 ,...,T ,
where
a
D
c
c
c
d
d
d
t
1t
v
lt
m+1,t
t
y
1t
m+1
it
=
i
=
=
t
=
=
=
=
,t
a
d
a
=
a
ot
u
d
=
1t
ot
lt
it
y
1t
it y
a
it-1
a
a
m,t-1
ot-1
d
u
l-1,t-1
d
1-1,t-1
it
m,t-1
[
[
l=2,...,m
l=2,...,m
]
]
-
8/7/2019 Okui Causality Between Economic and Political Freedom
14/38
14
(5)1
In the above equation, an individual effect disappears, and we can estimate theequation without considering the individual effect. And to avoid the simultaneous
problem, I use the following vector of instrumental variables as Holtz-Eakin, Newey,
and Rosen (1988, 1989) do:
Zt = [1, yt-2,...,y1, xt-2,...,x1]
Note that number of instrumental variables varies as the time period t changes. If
we can restrict that the lag coefficients are constant over time, the equation (5) can
be written:
D yit = at +
m
l=1
a l D yit-1 +
m
l=1
dl D xtt-1 + vit
(6).
Therefore in the method, it is important to check whether we can put restriction on
the equation (5).
Holtz-Eakin, Newey, and Rosen (1988, 1989) use the following estimation
procedure. i) Estimate the equation (5) for each time period using two-stage least
squares (2SLS), ii) using the residuals and the matrix of instruments, estimate the
joint covariance of the disturbance terms, iii) estimate all the parameters
simultaneously using generalized least squares on the stacked equations. This
procedure can be applied in the case where we put restriction on the equation (5).
The above procedure can be seen three-stage least squares (3SLS). And
their technique is GMM (Generalized Method of Moments Estimation) in that
instrumental variables are chosen so that they do not correlate with disturbance terms.
Hence I use 3SLS and GMM command of an econometric software package: TSP
when estimating the relevant equations. When I apply 3SLS and GMM, only the
commands 3SLS and GMM are different and the other part of the programs is the
same. Wooldridge (2002) refers to the difference between the GMM 3SLS andtraditional 3SLS. He says that GMM 3SLS is more rigorous. Therefore my
results of GMM are expected to be more reliable than the ones of 3SLS.
What we are the most interested in is causality between x and y. The
framework of the paper is convenient because we can test causality not estimating
the original parameters of the equation (4) but using the equation (5). 1=2==
m=0 in (4) means d1=d2==dm=0 in (5). Hence if the latter is rejected, the
former is rejected. Therefore if d1=d2==dm=0 is rejected, we can judge that "x
causesy." Moreover, if the lag coefficients are constant over time, we can check
-
8/7/2019 Okui Causality Between Economic and Political Freedom
15/38
15
1=2 ==m =0 directly by using equation (6).
I estimate the equation (5) or (6), and conduct causality tests between
political freedom and economic freedom. Following Holtz-Eakin, Newey, andRosen (1988, 1989), I take the procedure as follows.
1) I check whether parameters are stationary over time.
I choose a relatively large value ofm , estimate the autoregressions with and
without the restriction of parameters stationary, and compare the sum of squared
residuals. If the sum of squared residuals does not become large with the
restriction, we can judge that parameters are stationary.
2) I check what is the correct lag length, m.
Next, I check how many lags the autoregression has. I estimate
autoregression that has one-smaller lag length, compare its sum of squared residuals
with the one that is gotten in the process 1). If the sum of squared residuals does
not become large in the case of smaller lag length, we can judge that the smaller lag
length is accepted. Then we reduce the lag length and repeat the same test. If the
sum of squared residuals in the case of smaller lag length becomes large and the
hypothesis of the smaller lag length is rejected, we choose the lag length in the
previous process.
3) I check causality between political freedom and economic freedom.
Now we have the selected autoregressions. I estimate the autoregression
with and without the restriction of non-causality: d1=d2==dm=0 or1=2 ==
m = 0, and compare the sums of squared residuals. For example, if sum of
squared residuals becomes large with the restriction that economic freedom does not
cause political freedom, we can judge that economic freedom causes political
freedom.
7. Estimated Results
7.1. Political freedom equation
Let us start estimating a political freedom equation: an explained variable is political
freedom in the equation (5). I will estimate parameters for the last two years: 1995
and 2000 and start the case ofm=2. Equation (5) in the case ofm=2 has three lags.
That is,
-
8/7/2019 Okui Causality Between Economic and Political Freedom
16/38
16
D P1995 = a1995 + c1,1995 P1990 + c2,1995 P1985 + c3,1995 P1980 + d1,1995 E1990 + d2,1995 E1990 + d3,1995 E1980 +v1995
D P2000 = a2000 + c1,2000 P1995 + c2,2000 P1990 + c3,2000 P1985 + d1,2000 E1995 + d2,2000 E1990 + d3,2000 E1985 +v2000
( 7 ) ,
where Pt is political freedom of period t,Et is economic freedom of period t, and at
is constant of period t. For the 1995 equation, instrumental variables are political
freedom (P) of 1985, 1980, 1975 and economic freedom (E) of 1985, 1980, 1975
plus a constant term. For the 2000 equation, instrumental variables are P of 1990,
1985, 1980, 1975 andEof 1990, 1985, 1980, 1975 plus a constant term. I estimate
systems of these two periods equations by using three-stage least squares (3SLS).
The 3SLS estimation produced the sum of squared residuals Q: 1.721.
Next, let us put the restriction that all parameters are stationary. The equation on
that we put the restriction becomes
D Pt = a0 +a 1DPt-1+ a2D Pt-2 + d1DEt-1 + d 2,DEt-2 +v ( 8 ) .
I estimate the equation using the same instrumental variables as the case of equation
(7) and estimate it using 3SLS technique . As a result, I get the restricted sum of
squared residuals QR : 18.938.L= QR Q follows a chi-distribution with degrees of 8,
because
1,1990=1,1995=1,2000, 2,1990=2,1995=2,2000
1,1990=1,1995=1,2000, 2,1990=2,1995=2,2000
are restricted. Actually obtained L is 17.217, because L= QR Q =18.938
1.721=17.217. This value is the one of which the right side of the chi-distribution
is 2.8 percent, shown asp in Table 9. Therefore this hypothesis that all parameters
are stationary is rejected at the significance level 5 percent. Since the sum of
squared residuals increases very much when we put the restriction, we cannot accept
that all parameters are stationary.
Now, let us reduce lag length from m=2 and investigate how many lags ourselected model has. Obtained restricted sum of squared residuals QR is 9.027 when
we restrict m=1. Changing from m=2 to m=1 is the same as putting the restriction
that the coefficients of one lagged variable P and one lagged variable Eare zero for
each 1995 political freedom equation and 2000 political freedom equation. Hence
the number of restriction is 4. Therefore L, which is gotten by subtracting the sum
of squared residuals Q of the case m=2 from QR, follows 4 degrees of freedom.
Actually I get L= QR Q =9.027 1.721=7.038. This value is the one of which the
right side of the chi-distribution is 12.1 percent, shown as p. Therefore we can
-
8/7/2019 Okui Causality Between Economic and Political Freedom
17/38
17
accept m=1, although the restricted sum of squared residuals becomes quite large.
How about the case of m=0. Obtained restricted sum of squared residuals
QR is 29.483. Changing from m=1 to m=0 is the same as putting the restriction thatthe coefficients of one lagged variable P and one lagged variableEare zero for each
1995 political freedom equation and 2000 political freedom equation as well.
Hence the number of restrictions is 4. Actually obtained L= QR Q =29.483
0.027=16.367 is the value of which the right side of the chi-distribution is 0 percent,
shown asp. Therefore we strongly reject m=0. Sum of squared residuals increase
very much when m=0 is restricted.
Then let us conduct causality tests assuming that parameters are not
stationary and that lag length is 1. Putting the restriction of no causality from
economic freedom to political freedom is identical with that the coefficients of all
lagged variable E are zero for each 1995 political freedom equation and 2000
political freedom equation. Hence the number of restrictions is 4. Actually
obtained L= QR Q =14.723 9.027=5.696 is the value of which the right side of
the chi-distribution is 22.3 percent, shown as p. Therefore we accept the no
causality restriction. We cannot say that economic freedom causes political freedom.
Table 9 summaries the above procedure.
Table 9. Model selection of political freedom equation using 3SLS
Q L Degrees of Freedom p
) m =2 1.721 ----) all param eters stationary 18.938 17.217 8 0.028 **
) m =1 (given i) 9.027 7.038 4 0.121) ExcludeE (given i) 13.117 11.396 6 0.077 *
V ) m =0 (given iii) 29.483 16.367 4 0.000vi) ExcludeE (given i) 14.723 5.696 4 0.223
a) given m=2, we reject the hypothesis: all parameters are stationary.
b) given that all parameters are not stationary, we accept the hypothesis: m=1.c) given that all parameters are not stationary, we reject m=0.
d) given that all parameters are not stationary and that m=1, we accept the
hypothesis: no causality from economic freedom to political freedom.
In sum, the following model is selected: all parameters are not stationary,
m=1, and no causality from economic freedom to political freedom That is, we get
the result that economic freedom does not affect political freedom.
When we put restriction m=1 from m=2, restricted sum of squared residuals
-
8/7/2019 Okui Causality Between Economic and Political Freedom
18/38
18
becomes quite large. Hence we may see that hypothesis: m=1 is rejected. If it is
rejected, m=2 model is selected. Hence I conduct causality tests assuming m=2.
In this case, the hypothesis of no causality is neither rejected at the 1 percentsignificance level nor 5 percent significance level but 10 percent significance level.
In this case, we can see slight causality from economic freedom to political freedom.
I take the same procedure using GMM (Generalized Method of Moments
Estimation). Estimation by GMM is expected to be more rigorous than the one by
3SLS. I estimate parameters for the last two years: 1995 and 2000 and start the
case of m=2. Sum of squared residuals Q is 0.02026. Sum of squared residuals
QR when putting restriction that all parameters are stationary is 0.11227. L= QR
Q follows a chi-distribution with degrees of 8, and actually obtained L is 0.09201.
This value is the one of which the right side of the chi-distribution is almost 100
percent, shown as p in Table 10. Therefore this hypothesis that all parameters are
stationary is accepted. Next I put restriction of m=1 in addition to the restriction
that all parameters are stationary. Obtained restricted sum of squared residuals QR
is 0.24289. Changing from m=2 to m=1 is the same as putting the restriction that
the coefficients of one lagged variable P and one lagged variableEare zero for each
1995 political freedom equation and 2000 political freedom equation. Hence the
number of restriction is 2. Therefore L = QR Q follows 2 degrees of freedom.
Actually obtained L is 0.13062. This value is small enough for us to accept the
hypothesis: m=1. Then I put restriction of m=0 assuming that all parameters are
stationary. Sum of squared residuals does not increase even in the case of m=0.
Therefore we accept m=0. Table 10 shows summary of the procedure.
Table 10. Model selection of political freedom equation using GMM
Q L Degrees of Freedom p
) m =2 0.02026 ---- 2 0.98992) all param eters stationary 0.11227 0.09201 8 1.00000
) m =1 (given ii) 0.24289 0.13062 2 0.93678) m =0 (given ii) 0.47402 0.23113 2 0.89086
a) given m=2, we accept the hypothesis: all parameters are stationary.
b) given that all parameters are stationary, we accept the hypothesis: m=1.
c) given that all parameters are stationary, we accept m=0.
Such being the case, it follows from the results of estimating political
freedom equation that the model of m=0 is selected. That is, political freedom can
be explained neither by past political freedom nor by past economic freedom. It
-
8/7/2019 Okui Causality Between Economic and Political Freedom
19/38
19
means that whether democratization progress or not has nothing to do with how
degree democratization and marketization progressed in past times. Of course,
economic freedom does not cause political freedom.
7.2. Economic freedom equation
Here I make the same analysis of economic freedom equation as I did of political
equation. That is, I estimate the following system equations.
D E1995 = a1995 + c1,1995 E1990 + c2,1995 E1985 + c3,1995 E1980 + d1,1995 P1990 + d2,1995 P1990 + d3,1995 P1980 +v1995
D E2000 = a2000 + c1,2000 E1995 + c2,2000 E1990 + c3,2000 E1985 + d1,2000 P1995 + d2,2000 P1990 + d3,2000 P1985 +v2000
or
D Et = a 0 +a 1DEt-1+ a2D EE-2 + d1DPt-1 + d 2,DPt-2 +v
And I make model selection and causality tests. Results of estimation using 3SLS
technique are shown in Table.11.
Table 11. Model selection of economic freedom equation using 3SLS
Q L Degrees of Freedom p
) m =2 1.563 ----) all param eters stationary 21.222 19.659 8 0.012 **
) m =1 (given i) 8.957 7.395 4 0.116) Exclude P (given i) 17.125 15.562 6 0.016 **
V ) m =0 (given iii) 29.451 20.494 4 0.000 ***
vi) Exclude P (given iii 15.249 6.292 4 0.178
We can see the following from Table 11.
a) given m=2, we reject the hypothesis: all parameters are stationary.
b) given that all parameters are not stationary, we accept the hypothesis: m=1.
c) given that all parameters are not stationary, we reject m=0.d) given that all parameters are not stationary and that m=1, we accept the
hypothesis: no causality from economic freedom to political freedom.
When we put restriction m=1 from m=2, restricted sum of squared residuals
becomes quite large like the political freedom equation case. Hence I conduct
causality tests assuming m=2. In this case, the hypothesis of no causality is not
rejected at the 5 percent significance level. Comparing the result of Table 9, we
can reject the no causality hypothesis with higher probability. We can see from the
-
8/7/2019 Okui Causality Between Economic and Political Freedom
20/38
20
results that it is more probable that political freedom causes economic freedom than
vice versa.
Results of estimation using 3SLS technique are shown in Table 12.Table 12. Model selection of political freedom equation using GMM
Q L Degrees of Freedom p
) m =2 0.00120 ---- 2 0.99987) all param eters stationary 0.06273 0.06153 8 1.00000) m =1 (given ii) 0.06526 0.00253 2 0.99874) m =0 (given ii) 0.09733 0.03207 2 0.98409
We can see the following from Table 12.
a) given m=2, we accept the hypothesis: all parameters are stationary.
b) given that all parameters are stationary, we accept the hypothesis: m=1.
c) given that all parameters are stationary, we accept m=0.
The model of m=0 is also selected. That is, political freedom can be explained
neither by past political freedom nor by past economic freedom. Political freedom
does not cause economic freedom.
7.3. Summary of results of OLSQ, 3SLS and GMM
I conducted Granger causality tests in section 5. Estimation technique in
conducting Granger causality tests was ordinary least squares OLSQ. Now we
have the estimated results of political-economic equations and causality tests
between political freedom and economic freedom using three technique: Ordinary
least squares (OLSQ), three-stage least squares (3SLS), and GMM (Generalized
Method of Moments Estimation). Summary of the results is reported in Table 13.
Table 13. Summary of results of OLSQ, 3SLS and GMM
Notes. In OLSQ column, results of causality tests of the cases of m=3, m=2, and m=1 are shown. Stationary or not shows
results of the tests about the hypothesis: all parameters are stationary. Lag length shows results of the tests: how many lag
length estimated autoregreesions have. On causality, EP means economic freedom causes political freedom and EPmeans economic freedom does not cause political freedom, for example. Difference of Arrow thickness reflects
significance level of causality tests. Thicker arrow shows that no causality hypothesis is rejected at smaller rejection
region.
OLSQ 3SLS GMM
Political Freedom
Equation
Economic Freedom
Equation
Political Freedom
Equation
Economic Freedom
Equation
Political Freedom
Equation
Economic Freedom
Equation
m=3 E P P E stationary or not not sta tionary not tsat ionary stationary st ationary
m=2 E P P E lag length m=1 m=1 m=0 m=0
m=1 E P P E causaltiy E P P E E P P E
-
8/7/2019 Okui Causality Between Economic and Political Freedom
21/38
21
From the results of OLSQ, we can see strong causality from political
freedom to economic freedom and weak causality from economic freedom to
political freedom. Political freedom causes economic freedom, but it is difficult tosay that economic freedom causes political freedom. These results are consistent
with the fact that we find overviewing data in section 4.
But the above estimation technique have a weak point that individual effects
are neglected. 3SLS and GMM estimation technique was used to overcome the
weak point. 3SLS estimation selected non-stationary and m=1 model for both
political and economic freedom equations. And in causality tests using these
models, we cannot see valid causality between political freedom and economic
freedom.
The results of GMM estimation technique is exp ected to be more rigorous
than 3 SLS technique. GMM estimation selected stationary and m=0 model for
both political and economic freedom equations. That is, both political and
economic freedom are neither explained by past political freedom nor by past
economic freedom. It also means that there is no causality between political
freedom and economic freedom.
. Causality among components of political freedom and economic freedom
Here I check causality between political freedom and economic freedom using
different data. As I explained in section 4, my political freedom index is composed
of political rights index and civil liberties index issued by Freedom House. I use
each index and transform it the same way as in section 4. That is, each index are
standardized and prepared for 1975, 1980, 1985, 1990, 1995, and 2000 respectvely,
meaning higher values means having more freedom.
Economic freedom data issued by the Fraser Institute have five components: 1)
Size of Government, 2) Legal Structure and Security of Property Rights, 3) Access to
Sound Money, 4) Freedom to Trade Internationally, 5) Regulation of Credit, Labor, andBusiness. Size of Government indicates the extent to which countries rely on
individual choice and markets rather than the political process to allocate resources and
goods and services. Legal Structure and Security of Property Rights indicates the
extent to which countries have a legal system consistent with economic freedom, such
as rule of law, security of property rights, an independent judiciary, and an impartial
court system. Access to Sound Money indicates the extent to which countries are
successful in stabilizing value of money. Freedom to Trade Internationally indicates
the extent to which countries do not restrain free international exchange by using tariffs,
-
8/7/2019 Okui Causality Between Economic and Political Freedom
22/38
22
quotas, and so on. Regulation of Credit, Labor, and Business indicates the extent to
which countries do not restrain freedom of exchange in credit, labor, and product
markets by using control on interest rates, minimum wage, business entry regulation,and so on. I use each index and transform it the same way. That is, each index are
standardized and prepared for 1975, 1980, 1985, 1990, 1995, and 2000, meaning
higher values means having more freedom.
A List of components of political freedom and economic freedom is written
in Table 14. There are 10 combinations of political freedom and economic freedom.
I take the same procedure for each combination as taken in section 7. Table 15
shows summaries of results of OLSQ, 3SLS and GMM for the all combinations.
Details of the procedures are shown in Appendix.
Table 14. List of components of political freedom and economic freedom
Political Freedom Economic Freedom
Political Rights (PR)
Civil Liberties (CL)
Size of Government (SG)
Legal Structure and Security of Property Rights (LS)
Access to Sound Money (AS)
Freedom to Trade Internationally (FT)
Regulation of Credit, Labor, and Business. (RC)
Table 15. Summary of results of OLSQ, 3SLS and GMM
Causality between Political Rights (PR) and Size of Government (SG)
OLSQ 3SLS GMM
Political Freedom
E uat ion
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Causality between Civil Liberties (CL) and Size of Government (SG)
OLSQ 3SLS GMM
Political Freedom
Equation
Economic Freedom
Equation
Political Freedom
Equation
Economic Freedom
Equation
Political Freedom
Equation
Economic Freedom
Equation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
-
8/7/2019 Okui Causality Between Economic and Political Freedom
23/38
23
Causality between Political Rights (PR) and Legal Structure & Security of Property Rights (SP)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Causality between Civil Liberties (CL) and Legal Structure & Security of Property Rights (SP)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Causality between Political Rights (PR) and Access to Sound Money (AS)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Causality between Civil Liberties (CL) and Access to Sound Money (AS)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
-
8/7/2019 Okui Causality Between Economic and Political Freedom
24/38
24
Causality between Political Rights (PR) and Freedom to Trade Internationally (FT)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Causality between Civil Liberties (CL) and Freedom to Trade Internationally (FT)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Causality between Political Rights (PR) and Regulation of Credit, Labor & Banking (RC)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Causality betweenCivil Liberties (CL) and Regulation of Credit, Labor & Banking (RC)
OLSQ 3SLS GMM
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
Political Freedom
E uation
Economic Freedom
E uation
m =3 E P P E stationary or not not stationary not tsationary stationary stationary
m =2 E P P E lag length m =1 m =1 m =0 m =0
m =1 E P P E causaltiy E P P E E P P E
Using Table 15, I count the number of rejecting no causality hypothesis at 5 percent
re and 1 percent rejection region for each OLSQ and 3SLS and GMM. Table 16
shows the result. When I use OLSQ and 3SLS technique, I find some cases where
-
8/7/2019 Okui Causality Between Economic and Political Freedom
25/38
25
economic freedom causes political freedom and political freedom causes economic
freedom. There are more cases where political freedom causes economic freedom
than the ones where economic freedom causes political freedom. It follows from itthat political freedom are more likely to cause economic freedom than vice versa.
There are not so many cases where economic freedom causes political freedom.
And the probability that political freedom causes economic freedom is higher than
the one that economic freedom causes political freedom. These facts are also
consistent with the ones that I find in section 4.
When I use GMM technique, on the contrary, no causality is found. OLSQ
has a weak point that individual effects are neglected. Hence I used 3SLS and
GMM to improve the precision of estimation. Because GMM technique gives more
accurate estimated results, the results of GMM are more reliable than the ones of
3SLS. But the results of GMM do not see any causality. Rigorous analysis shows
that there is no causality between political freedom and economic freedom.
Table 16. The number of cases where causality is accepted strongly.
OLSQ 3SLS GMM
E P
5
out of 30
P E
18
out of 30
E P
3
out of 10
P E
5
out of 10
E P
0
out of 10
P E
0
out of 30
8. Conclusion
The aim of this paper was to investigate causality between political freedom
and economic freedom. Since economic freedom data do not have long time series,
I conducted causality tests using panel data that have many countries. Some
previous works conducted causality tests without considering individual effects.
Hence I used the model that has individual effects. And I also consider the
possibility that coefficient parameters of political and economic freedom are
stationary and that lag length changes by using the method by Holtz-Eakin, Newey,
and Rosen (1988, 1989).
If we overview political freedom data and economic freedom data, in
many countries economic liberalization progressed after democratization had been
developed. Granger causality tests that do not take account of individual effects
support that democratization promotes marketization.
But the estimation that takes account of individual effects leads to the
-
8/7/2019 Okui Causality Between Economic and Political Freedom
26/38
26
following result. Political freedom does not cause economic freedom. And,
economic freedom does not cause political freedom. That is, neither causality from
democratization to marketization nor causality from marketization todemocratization is found.
A glance at the data shows that democratization promotes marketization.
But more rigorous analysis says that there is no causality between political freedom
and economic freedom. This is the conclusion of the paper.
What policy implication is lead by the conclusion? At the beginning of
the paper, we mentioned that the result of the paper is considered to contribute to
solving the question which should be put priority on between democratization and
marketization. But it is with regret that this disappointing conclusion cannot lead
the answer. We may, therefore, confirm that we should be careful about which we
should put priority on when we think about development strategy of developing
countries. In any event, in a state where many researches that explore the
relationship between institutions and economic performance has come out,
researches that explore the relationship between institutions themselves, such as the
relationship between political freedom and economic freedom, seem to be weak.
The research of this field seems to be more important in the future.
Notes
1. Holtz-Eakin, Newey, and Rosen (1988, 1989) consider the parameter multiplying
the individual effect, which allows us to analyze the case where the individual effect
is stationary, but the other parameters are not. Because we are not interested in
whether the individual effect is stationary or not, I omit the parameter. That is,
the equation here is the special case of the one in Holtz-Eakin, Newey, and Rosen.
-
8/7/2019 Okui Causality Between Economic and Political Freedom
27/38
27
References
Ali, A. and Crain, M. (2001). Political Regimes, Economic Freedom, Institutions and Growth.
Journal of Public Finance and Public Choice 19: 3-21.Barro, R. J. (1996). Democracy and Growth. Journal of Economic Growth 1: 1-27.
Chong,A. and Calderon, C. (2000). Causality and Feedback between Institutional Measures and
Economic Growth. Economics and Politics 12: 69-81.
Dawson, J. W. (1998) . Institutions, Investment, and Growth: New Cross-Country and Panel Data
Evidence. Economic Inquiry 36: 603-619.
Dawson, J. W. (2003). Causality in the freedom-growth relationship. European Journal of
Political Economy 19: 479-495.
De Haan, J. and Siermann, C. (1995). New evidence on the relationship between democracy and
economic growth. Public Choice 86: 175-98.
De Haan, J. and Siermann, C. (1998). Further evidence on the relationship between democracy
and economic growth. Public Choice 95: 363-380.
De Haan, J. and Sturm, J. (2003) . Does more democracy lead to greater economic freedom? New
evidence for developing countries. European Journal of Political Economy 19: 547-563.
Dethier, J., Ghanem, H. and Zoli, E. (1999) . Does Democracy Facilitate the Economic Transition?
An Empirical Study of Central and Eastern Europe and the Former Soviet Union.
Journal for Institutional Innovation, Development and Transition 3: 15-30.
Durham, B. (1999). Economic Growth and Political Regimes. Journal of Economic Growth 4:
81-111.
Eaton, S. T. and Walker, M.A. (1997). Income, Growth, and Economic Freedom.
American Economic Review Papers and Proceedings 87: 328-332.
Farr, W. K., Lord, R.A. and Wolfenbarger, J. L. (1998) . Economic Freedom, Political Freedom,
and Economic Well-Being: A Causality Analysis. Cato Journal 18: 247-262.
Feng, Y. (2003) . Democracy, Governance, and Economic Performance, Cambridge: The MIT
Press.
Freedom House, FH Country Ratings, All Countries' Scores.
[http://freedomhouse.org/ratings/index.htm].
Friedman, M. (1962). Capitalism and Freedom, Chicago: University of Chicago Press.
Gwartney, James, Robert Lawson, et al. (2002). Economic Freedom of the World: 2002 Annual
Report. Vancouver: The Fraser Institute. [http://www.freetheworld.com.]
Holtz-Eakin, D.,Newey, W. and Resen, H. (1988). Estimating Vector Autoregressions with Panel
Data. Econometrica 56: 1371-1395.
Holtz-Eakin, D.,Newey, W. and Resen, H. (1989). The Revenue-Expenditures Nexus: Evidence
from Local Government Data. International Economic Review 30: 415-429.
-
8/7/2019 Okui Causality Between Economic and Political Freedom
28/38
28
Knack, S. and Keefer, P. (1995). Institutions and Economic Performance: Cross-Country Tests
Using Alternative Institutional Measures. Economics and Politics 7: 207-227.
Lal, D. and Myint, H. (1996) . The Political Economy of Poverty, Equity, and Growth. Oxford:Oxford University Press.
Sachs, J. and Woo, W. (1994). Structural factors in the economic reforms of China, Eastern
Europe, and the Former Soviet Union. Economic Policy 94: 102-145.
Scully, W. and Slottje, J. (1991). Ranking Economic Liberty across Countries . Public
Choice 69: 121-152.
Sen, A. (1999) . Development as Freedom. New York: Alfred A. Knopf.
Vanssay, X. D. and Spindler, Z.A. (1994). Freedom and Growth: Do Constitutions matter?.
Public Choice 78: 359-372.
Wooldridge, J. (2002) Econometric Analysis of Cross Section and Panel Data. Cambridge: The
MIT Press.
Wu, W. and Davis, O. (1999). The Two Freedoms, economic growth and development: An
Empirical Study. Public Choice 100: 39-64.
-
8/7/2019 Okui Causality Between Economic and Political Freedom
29/38
29
Appendix .
Causality between Political Rights (PR) and Size of Government (SG)
1) OLSQPolitical Freedom Equation
Sum of
squared
residuals
FDegrees of
Freedom
m =3 42.18 -----
Exclude 42.78 0.96 3,203 0.412
m =2 42.83 -----
Exclude 43.37 1.29 2,205 0.278
m =1 42.86 -----
Exclude 43.44 2.80 1,207 0.096 *
Table Econom ic Freedom EquaSum of
squared
residuals
FDegrees of
Freedom
m =3 41.22 -----
Exclude 42.06 1.37 3,203 0.253
m =2 41.34 -----
Exclude 42.16 2.04 2,205 0.133
m =1 41.58 -----
Exclude 42.17 2.92 1,207 0.089 *
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 3.733 ---- 2 0.423) all param eters stationary 30.991 27.258 8 0.001 **
) m =1 (given i) 4.712 0.979 4 0.913) m =0 (given iii) 43.784 39.071 4 0.000 ***
V ) ExcludeE (given iii 11.711 6.999 4 0.136
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.424 ---- 2 0.458) all param eters stationary 73.708 73.284 8 0.000 ***
) m =1 (given i) 3.076 2.651 4 0.618) m =0 (given iii) 109.878 0.000 4 0.003 ***
V ) Exclude P (given iii 8.427 5.351 4 0.253
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.03793 ----) all param eters stationary 0.25746 0.21954 8 0.99999) m =1 (given ii) 0.27197 0.01451 2 0.99277) m =0 (given ii) 0.27883 0.00685 2 0.99658
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.02851 ----) all param eters stationary 0.04662 0.01810 8 1.00000) m =1 (given ii) 0.05818 0.01156 2 0.98961) m =0 (given ii) 0.32268 0.26450 2 0.94808
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
30/38
30
Causality between Civil Liberties (CL) and Size of Government (SG)
1) OLSQPolitical Freedom Equation
Sum of
squared
residualsF
Degrees of
Freedom
m =3 28.95 -----
ExcludeE 29.14 0.46 3,203 0.713
m =2 28.97 -----
ExcludeE 29.17 0.71 2,205 0.493
m =1 29.24 -----
ExcludeE 29.31 0.50 1,207 0.479
Econom ic Freedom EquationSum of
squared
residualsF
Degrees of
Freedom
m =3 40.67 -----
Exclude P 42.06 2.30 3,203 0.078 *
m =2 41.08 -----
Exclude P 42.16 2.70 2,205 0.069
m =1 41.27 -----
Exclude P 42.17 4.51 1,207 0.035 **
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 2.657 ----) all param eters stationary 79.953 77.296 8 0.000 ***
) m =1 (given i) 6.572 3.915 4 0.418) m =0 (given iii) 150.311 143.739 4 0.000 ***
V ) ExcludeE (given iii 10.713 4.141 4 0.387
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 1.069 ----) all param eters stationary 115.087 114.018 8 0.000 ***
) m =1 (given i) 4.825 3.757 4 0.440) m =0 (given iii) 234.178 0.000 4 0.000 ***
V ) Exclude P (given iii 11.074 6.249 4 0.181
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.01669 ----) all param eters stationary 0.10738 0.09069 8 1.00000) m =1 (given ii) 0.15038 0.04300 2 0.97873) m =0 (given ii) 0.19729 0.04691 2 0.97682
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.02851 ----) all param eters stationary 0.06969 0.04118 8 1.00000) m =1 (given ii) 0.19808 0.12839 2 0.98961) m =0 (given ii) 0.44989 0.25181 2 0.88170
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
31/38
31
Causality between Political Rights (PR) and Legal Structure & Security of Property
Rights (SP)
1) OLSQPolitical Freedom Equation
Sum of
squared
residualsF
Degrees of
Freedom
m =3 12.64 -----
ExcludeE 13.59 2.34 3,93 0.079 *
m =2 12.70 -----
ExcludeE 13.94 4.64 2,95 0.012 **
m =1 12.94 -----
ExcludeE 13.95 7.58 1,97 0.007 ***
Econom ic Freedom EquationSum of
squared
residualsF
Degrees of
Freedom
m =3 23.50 -----
Exclude P 24.90 1.84 3,93 0.145
m =2 24.43 -----
Exclude P 25.38 1.85 2,95 0.163
m =1 25.20 -----
Exclude P 26.12 3.55 1,97 0.062 *
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 6.114 ----) all param eters stationary 28.811 22.697 8 0.004 ***
) m =1 (given i) 10.473 4.360 4 0.360) m =0 (given iii) 34.497 24.023 4 0.001 ***
V ) ExcludeE (given iii 29.214 18.741 4 0.001 ***
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 8.280 ----) all param eters stationary 51.350 43.069 8 0.000 ***
) m =1 (given i) 10.104 1.824 4 0.768) m =0 (given iii) 59.341 49.237 4 0.000 ***
V ) Exclude P (given iii 12.249 2.145 4 0.709
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.05932 ----) all param eters stationary 0.24708 0.18777 8 1.00000) m =1 (given ii) 0.28428 0.03720 2 0.97873) m =0 (given ii) 0.55054 0.26626 2 0.87535
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.04276 ----) all param eters stationary 0.11991 0.07715 8 1.00000) m =1 (given ii) 0.16445 0.04454 2 0.97798) m =0 (given ii) 0.31288 0.14843 2 0.92847
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
32/38
32
Causality between Political Rights (CL) and Legal Structure & Security of Property
Rights (SP)
1) OLSQPolitical Freedom Equation
Sum of
squared
residualsF
Degrees of
Freedom
m =3 10.20 -----
ExcludeE 12.14 5.89 3,93 0.001 ***
m =2 10.72 -----
ExcludeE 12.19 6.49 2,95 0.002 ***
m =1 10.76 -----
ExcludeE 12.22 7.58 1,97 0.000 ***
Econom ic Freedom EquationSum of
squared
residualsF
Degrees of
Freedom
m =3 22.98 -----
Exclude P 24.90 2.58 3,93 0.058 *
m =2 23.68 -----
Exclude P 25.38 3.41 2,95 0.037 **
m =1 23.93 -----
Exclude P 26.12 8.87 1,97 0.004 ***
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 2.715 ----) all param eters stationary 58.563 55.848 8 0.000 ***
) m =1 (given i) 25.538 22.823 4 0.001 ***
) ExcludeE (given i) 27.318 24.604 6 0.000 ***
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 10.308 ----) all param eters stationary 44.448 34.140 8 0.000 ***
) m =1 (given i) 21.789 11.480 4 0.022 **
) m =0 (given iii) 53.958 32.169 4 0.000 ***
V ) Exclude P (given iii 29.539 7.751 4 0.101
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.01167 ----) all param eters stationary 0.31318 0.30151 8 0.99980
) m =1 (given ii) 0.48797 0.17479 2 0.91632) m =0 (given ii) 0.63625 0.14828 2 0.92854
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.07389 ----) all param eters stationary 0.45140 0.37750 8 0.99995) m =1 (given ii) 0.65691 0.20551 2 0.90235) m =0 (given ii) 0.83340 0.17649 2 0.61554
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
33/38
33
Causality between Political Rights (PR) and Access to Sound Money (AS)
1) OLSQ
Political Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 44.02 -----
ExcludeE 45.15 1.84 3,215 0.140
m =2 44.62 -----
ExcludeE 45.70 2.62 2,217 0.075 *
m =1 45.62 -----
ExcludeE 45.91 1.41 1,219 0.236
Econom ic Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 95.77 -----
Exclude P 108.33 9.40 3,215 0.000 ***
m =2 96.42 -----
Exclude P 108.66 13.78 2,217 0.000 ***
m =1 98.29 -----
Exclude P 111.51 29.44 1,219 0.000 ***
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 5.462 ----) all param eters stationary 33.847 28.385 8 0.000 ***
) m =1 (given i) 10.128 4.666 4 0.323) m =0 (given iii) 45.422 35.293 4 0.000 ***
V ) ExcludeE (given iii 14.183 4.054 6 0.399
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 4.996 ----) all param eters stationary 103.590 98.594 8 0.000 ***
) m =1 (given i) 7.789 2.792 4 0.593) m =0 (given iii) 169.095 161.306 4 0.000 ***
V ) Exclude P (given iii 44.531 36.743 4 0.000 ***
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.03869 ----) all param eters stationary 0.24812 0.20943 8 1.00000
) m =1 (given ii) 0.32114 0.07303 2 0.96415) m =0 (given ii) 0.34739 0.02625 2 0.98696
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.02083 ----) all param eters stationary 0.36938 0.34855 8 0.99997) m =1 (given ii) 0.57268 0.20330 2 0.90335) m =0 (given ii) 1.04683 0.47415 2 0.78893
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
34/38
34
Causality between Political Rights (CL) and Access to Sound Money (AS)
1) OLSQ
Political Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 32.52 -----
ExcludeE 32.73 0.46 3,215 0.712
m =2 32.56 -----
ExcludeE 32.78 0.73 2,217 0.483
m =1 33.01 -----
ExcludeE 33.08 0.45 1,219 0.501
Econom ic Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 93.23 -----
Exclude P 108.33 11.61 3,215 0.000 ***
m =2 96.46 -----
Exclude P 108.66 16.31 2,217 0.000 ***
m =1 97.79 -----
Exclude P 111.51 30.72 1,219 0.000 ***
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 1.129 ----) all param eters stationary 19.759 18.630 8 0.017 **
) m =1 (given i) 2.954 1.826 4 0.768) m =0 (given iii) 27.717 24.762 4 0.000 ***
V ) ExcludeE (given iii 3.699 0.744 6 0.946
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 5.008 ----) all param eters stationary 140.423 135.415 8 0.000 ***
) m =1 (given i) 15.428 10.420 4 0.033 **
V ) Exclude P (given iii 54.557 39.129 6 0.000 ***
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.00464 ----) all param eters stationary 0.07064 0.06600 8 1.00000) m =1 (given ii) 0.16392 0.09329 2 0.95443
) m =0 (given ii) 0.24262 0.07870 2 0.96142 Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.03381 ----) all param eters stationary 0.37664 0.34283 8 0.99997) m =1 (given ii) 0.40631 0.02967 2 0.98527) m =0 (given ii) 2.51319 2.10688 2 0.34875
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
35/38
35
Causality between Political Rights (PR) and Freedom to Trade Internationally (FT)
1) OLSQ
Political Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 31.11 -----
ExcludeE 32.29 2.14 3,169 0.097 *
m =2 31.80 -----
ExcludeE 32.77 2.62 2,171 0.076 *
m =1 32.12 -----
ExcludeE 32.80 3.67 1,173 0.057 *
Econom ic Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 34.81 -----
Exclude P 36.51 2.75 3,169 0.044 **
m =2 34.83 -----
Exclude P 36.58 4.29 2,171 0.015 **
m =1 35.47 -----
Exclude P 37.51 9.95 1,173 0.002 ***
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 1.431 ----) all param eters stationary 50.539 49.108 8 0.000 ***
) m =1 (given i) 4.044 2.613 4 0.625) m =0 (given iii) 66.645 62.601 4 0.000 ***
V ) ExcludeE (given iii 15.118 11.074 4 0.026 **
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 4.016 ----) all param eters stationary 66.184 62.168 8 0.000 ***
) m =1 (given i) 5.438 1.422 4 0.840) m =0 (given iii) 94.831 89.393 4 0.000 ***
V ) Exclude P (given iii 16.532 11.094 4 0.026 **
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.00568 ----) all param eters stationary 0.41376 0.40808 8 0.99994
) m =1 (given ii) 0.44643 0.03267 2 0.98380) m =0 (given ii) 0.52331 0.07688 2 0.96229
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.01913 ----) all param eters stationary 0.69601 0.67688 8 0.99958) m =1 (given ii) 0.70754 0.01153 2 0.99425) m =0 (given ii) 1.35121 0.64367 2 0.72482
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
36/38
36
Causality between Civil Liberties (CL) and Freedom to Trade Internationally (FT)
1) OLSQ
Political Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 21.72 -----
ExcludeE 22.67 2.47 3,169 0.064 *
m =2 22.55 -----
ExcludeE 22.67 0.47 2,171 0.624
m =1 22.63 -----
ExcludeE 22.73 3.67 1,173 0.381
Econom ic Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 36.23 -----
Exclude P 36.51 0.44 3,169 0.725
m =2 36.29 -----
Exclude P 36.58 0.67 2,171 0.515
m =1 37.24 -----
Exclude P 37.51 9.95 1,173 0.260
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 1.771 ----) all param eters stationary 18.730 16.959 8 0.031 **
) m =1 (given i) 6.439 4.668 4 0.323) m =0 (given iii) 37.455 31.016 4 0.000 ***
V ) ExcludeE (given iii 9.345 2.907 4 0.574
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.432 ----) all param eters stationary 52.446 52.013 8 0.000 ***
) m =1 (given i) 2.816 2.383 4 0.666) m =0 (given iii) 66.102 63.286 4 0.000 ***
V ) Exclude P (given iii 6.425 3.609 4 0.461
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.01630 ----) all param eters stationary 0.14032 0.12403 8 1.00000
) m =1 (given ii) 0.29842 0.15810 2 0.92399) m =0 (given ii) 0.45569 0.15727 2 0.92438
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.00117 ----) all param eters stationary 0.41652 0.41535 8 0.99993) m =1 (given ii) 0.41920 0.00269 2 0.99866) m =0 (given ii) 0.56285 0.14365 2 0.93069
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
37/38
37
Causality between Political Rights (PR) and Regulation of Credit, Labor & Banking
(RC)
1) OLSQPolitical Freedom Equation
Sum of
squared
residualsF
Degrees of
Freedom
m =3 27.02 -----
ExcludeE 27.23 0.35 3,139 0.792
m =2 27.50 -----
ExcludeE 27.53 0.08 2,141 0.926
m =1 27.56 -----
ExcludeE 27.57 0.07 1,143 0.791
Econom ic Freedom EquationSum of
squared
residualsF
Degrees of
Freedom
m =3 35.94 -----
Exclude P 40.30 5.62 3,139 0.001 ***
m =2 35.99 -----
Exclude P 40.30 8.46 2,141 0.000 ***
m =1 36.31 -----
Exclude P 40.41 16.16 1,143 0.000 ***
2) 3SLS
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 2.718 ----) all param eters stationary 67.802 65.084 8 0.000 ***
) m =1 (given i) 9.627 6.908 4 0.141) m =0 (given iii) 114.167 104.540 4 0.000 ***
V ) ExcludeE (given iii 12.530 2.904 4 0.574
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 5.437 ----) all param eters stationary 55.077 49.640 8 0.000 ***
) m =1 (given i) 7.715 2.277 4 0.685) m =0 (given iii) 66.158 58.443 4 0.000 ***
V ) Exclude P (given iii 19.883 12.168 4 0.016 **
3) GMM
Political Freedom Equation
Q L Degrees of Freedom p
) m =2 0.00775 ----) all param eters stationary 0.17197 0.16422 8 1.00000) m =1 (given ii) 0.18606 0.01409 2 0.99298) m =0 (given ii) 0.34376 0.15769 2 0.92418
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.00557 ----) all param eters stationary 0.25987 0.25429 8 0.99999) m =1 (given ii) 0.27371 0.01384 2 0.99310) m =0 (given ii) 0.29797 0.02426 2 0.98794
*** significant at the 1-percent level ** Significant at the 5-percent level * Significant at the 10-percent level
-
8/7/2019 Okui Causality Between Economic and Political Freedom
38/38
Civil Liberties (CL) and Regulation of Credit, Labor & Banking (RC)
1) OLSQ
Political Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 18.88 -----
ExcludeE 19.14 0.64 3,139 0.589
m =2 19.01 -----
ExcludeE 19.15 0.53 2,141 0.592
m =1 19.43 -----
ExcludeE 19.48 0.44 1,143 0.510
Econom ic Freedom EquationSum ofsquared
residualsF
Degrees of
Freedom
m =3 33.68 -----
Exclude P 40.30 9.11 3,139 0.000 ***
m =2 33.86 -----
Exclude P 40.30 13.42 2,141 0.000 ***
m =1 34.07 -----
Exclude P 40.41 26.61 1,143 0.000 ***
2) 3SLS
P olitical Freedom Equation
Q L Degrees of Freedom p
) m=2 1.809 ----) all param eters stationary 15.885 14.076 8 0.080 *) m=1 (given i) 3.283 1.474 4 0.831) m=0 (given iii) 25.902 22.619 4 0.000 ***
V) ExcludeE (given iii) 6.369 3.086 4 0.544
vi) m=1 (given ii) 20.348 4.463 4 0.347(vii) m=0 (given iii) 25.902 5.554 4 0.235
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 5.429 ----) all param eters stationary 60.371 54.942 8 0.000 ***
) m =1 (given i) 6.718 1.289 4 0.863) m =0 (given iii) 76.750 70.033 4 0.000 ***
V ) Exclude P (given iii 23.390 16.672 4 0.002 ***
3) GMM
Political Freedom Equation
Q LDegrees of Freedom
p) m =2 0.02092 ----) all param eters stationary 0.14624 0.12533 8 1.00000) m =1 (given ii) 0.30297 0.15673 2 0.92463) m =0 (given ii) 0.33071 0.02774 2 0.98623
Econom ic Freedom Equation
Q L Degrees of Freedom p
) m =2 0.00706 ----) all param eters stationary 0.24232 0.23526 8 0.99999) m =1 (given ii) 0.29992 0.05759 2 0.97161) m =0 (given ii) 0.36689 0.06697 2 0.96707