Political Instability, Freedom, and Economic Growth: Some Further Evidence

Author(s): Jakob de Haan and Clemens L. J. Siermann

Source: Economic Development and Cultural Change, Vol. 44, No. 2 (Jan., 1996), pp. 339-350
Published by: The University of Chicago Press
Stable URL: https://www.jstor.org/stable/1154407
Accessed: 25-01-2019 08:58 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1154407?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and
extend access to Economic Development and Cultural Change

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
Political Instability, Freedom, and Economic
Growth: Some Further Evidence*

Jakob de Haan and Clemens L. J. Siermann

University of Groningen, The Netherlands

I. Introduction
Over time, the impact of both political and institutional factors on th
rate of economic growth has received a lot of attention in the econom
literature. Since Adam Smith (if not before) it has been observed tha
economic performance depends in part on political and institutional
factors. Unfortunately, attempts to discover how and under what con
ditions such factors impinge on economic performance often did no
progress beyond anecdote.1 It is only recently that more systematic
attempts were undertaken to analyze the importance of political an
institutional factors in explaining cross-country variances in eco
nomic growth. For instance, many researchers have examined whethe
the democratic character of national political procedures and institu
tions can help explain cross-country differences in national develop
ment. L. Sirowy and A. Inkeles reviewed 13 studies, all of which
attempted to evaluate the economic consequences of variation in the
democratic character of national political regimes. Sirowy and Inkeles
conclude that the studies they considered present a very mixed and
confusing picture with regard to the effect of democracy on economi
growth.2
The relationship between civil and political liberty, on the one
hand, and economic growth, on the other, has also been the subject
of much discussion. Two conflicting hypotheses have been advanced
concerning the connection between economic growth and political
freedom. Some economists argue that freedom fosters economic per-
formance and hence economic growth; others pose that high growth
rates require economic controls and reduced freedom. Four recent
cross-section studies on economic growth have found evidence that
lack of civil and political liberties is negatively correlated with eco-
nomic growth.3 All of these studies used R. D. Gastil's measures of
civil liberties or political rights.4 G. W. Scully concludes that, for the
? 1996 by The University of Chicago. All rights reserved.
0013-0079/96/4402-0007$01.00

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
340 Economic Development and Cultural Change

115 market economies studied over the period 1960-80, the po

open societies that subscribe to political, civil, and economic
grow at three times the rate of societies in which these freed
restricted. For their group of 47 countries, R. C. Kormendi a
Meguire find that countries in the high civil liberty category exp
about 1% greater annual economic growth. Similarly, K. B. G
G. Tullock conclude that repressive countries in Africa and th
cas have about a 1.5-percentage-point lower annual growth r
do other countries included in their study. R. J. Barro's resu
sample of 98 countries also indicate that restricted political r
associated with lower per capita growth.
Recently, also the issue of political instability and eco
growth has been investigated by a number of scholars. Most
studies point to political instability as an important hindrance
nomic growth, since political instability reduces the supply
capital and labor. Investment is discouraged due to the incre
of capital loss, and political turmoil causes capital flight a
drain. Political unrest also hampers the establishment of
rights, which are necessary in order to realize productivity g
ciated with impersonal exchange.
In his cross-section model for 98 countries for 1960-85 Barro in-
cluded two variables to measure political instability: the number of
revolutions and coups per year and the number, per million population,
of political assassinations per year. He interprets these variables as
adverse influences on property rights and finds that both variables
exert a significant negative influence on economic growth. A. K. Fosu
examined the importance of political instability from 1960 to 1986 for
31 sub-Saharan African countries. He also concludes that political in-
stability has, on average, a deleterious impact on economic growth.6
It is very surprising that most studies reviewed so far do not differ-
entiate between different groups of countries. The first purpose of this
article is to examine whether empirical relationships between political
instability and political freedom-as measured globally by most other
researchers-are also valid for smaller geographic areas. The second
purpose is to analyze whether results of previous estimates are sensi-
tive with respect to the definition of measures for political instability
and freedom. We examine a sample of 97 countries from 1963 to 1988
to determine whether lack of political stability and lack of political
freedom are negatively correlated with economic growth. We find
some evidence that lack of political stability affects economic growth
in Africa. However, this finding is not robust, since other measures of
political instability alter this outcome quite dramatically. There is some
evidence that in Latin America political repression is negatively corre-
lated with economic development. However, we show that this conclu-

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
Jakob de Haan and Clemens L. J. Siermann 341

sion is highly dependent on the way the political

is constructed.
Kormendi and Meguire argue that political repression affects eco-
nomic growth mainly through the investment-income ratio (IIR). The
third purpose of this article is to extend their line of research by exam-
ining the relationship between political stability and political freedom
and the IIR. We find that in Africa and Asia our indicators for political
instability are negatively related to the IIR. There is some mixed evi-
dence that in Latin America political repression may hamper capital
formation.
The article is organized as follows. Section II discusses the model
and our data. Section III presents estimation results concerning the
relationship between political instability and economic growth. Section
IV contains our findings with respect to the impact of political freedom.
In this section we also address Barro's finding that when political insta-
bility and political freedom are simultaneously introduced into a
growth regression, the civil liberties measure loses its significance.
Finally, Section V summarizes the article and presents suggestions for
future research.

II. Our Model and the Data

We have estimated variants of the following cross-section equation

Q = ao + a, L + a2K + a3P+v, (1)

where Q, L, and K are the average growth rates of r

tion, and capital, respectively, and P is a measure of p
or lack of political freedom and v is the error term. T
and a2 are expected to be positive, while most previo
suggest that a3 will prove to be negative.
For the period 1963-88 we have constructed a
countries. This sample excludes the major oil-expo
These countries tend to have a high level of per capit
however, primarily caused by large amounts of inco
off natural resources. The mean annual growth rate
taken from the 1990 and 1991 issues of the Internat
Fund's International Financial Statistics (IFS) Yearboo
growth rate of capital, following Fosu, measured as m
tic investment as a percentage of GDP, is taken from
A. Heston's data set.9 Data on population growth
from this source.
Our measure of political instability (TRNS) is based on the total
number of government changes. This variable is assigned a value of
one if the number of government transfers exceeds seven (the median

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
342 Economic Development and Cultural Change

number of transfers for the countries in our sample) and zer

wise. In contrast to most other studies employing government tra
as indicator for political instability-see, e.g., S. Edwards and
bellini-we do not use the total number of government trans
such.1o We have also experimented with other measures of p
instability, taking six government transfers (the average fo
and eight transfers (the average for Latin America) as alternative
points. Data on government transfers are taken from C. L. Ta
D. Jodice for the period 1963-77.11 For 1978-88 we construct
instability measures on the basis of various issues of A. S.
Political Handbook of the World.12
Gastil has constructed two measures of freedom: civil liber
political liberty. The civil rights rankings purport to measure the
of the individual (e.g., independence of the judiciary and fre
the press). The political rights rankings are based on the deg
which individuals in a state have control over those who gove
measure is based on issues like freedom of the electoral proces
ence of intimidating violence, and the presence of an effective
tion. The index runs from one (most free) to seven (least fre
both measures of freedom are highly correlated, in Section IV
report the regression outcomes using dummy variables that a
on Gastil's political rights index. The dummy DPRI3 is on
average of Gastil's political rights index for the period 1973-86
or higher and it is zero otherwise. For the dummy DPRI4 the
of four is used as a cutoff point. By using two dummies we
to investigate the sensitivity of our results for alternative tra
tions. In Section IV we show that the outcomes are indeed sensitive
with respect to the construction of the political rights dummy.13
Table 1 summarizes our data. It is interesting to note that the
variance of our political instability measure is very similar for the
various groups of countries that we examine. Except perhaps for Af-
rica, this also holds true for our measures of political freedom. The
lower part of table 1 shows the partial correlation between our mea-
sures for political instability and political freedom. It follows that this
correlation is quite low.

III. Political Instability and Economic Growth

Before estimating the model, we first have to deal with a problem
that was identified by Kormendi and Meguire.14 They argue that if
the political-institutional variable affects economic development solely
through the investment ratio, inclusion of capital growth should
remove its effect. Therefore, we followed Kormendi and Me-
guire by excluding investment in estimating the model. As sug-
gested by these authors, we also estimated simple models for the IIR.

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
Jakob de Haan and Clemens L. J. Siermann 343

TABLE 1

SUMMARY MEASURES OF DEPENDENT AND INDEPENDENT VARIABLES

POLITICAL POLITICAL
STABILITY FREEDOM

Q K L TRNS DPRI3 DPRI4

Average:
World (97) 4.0 18.3 2.1 .52 .54 .47
Africa (37) 3.9 13.2 2.8 .30 .84 .76
Latin America (23) 3.7 16.6 2.3 .52 .56 .43
Asia (22) 6.2 20.6 3.0 .59 .68 .68
Standard deviation:
World 1.9 8.8 1.0 .50 .50 .50
Africa 2.2 8.4 .5 .46 .37 .43
Latin America 1.6 5.6 .8 .51 .51 .51
Asia 3.7 9.5 1.3 .50 .48 .48

DPRI3 and TRNS DPRI4 and TRNS

Partial correlation be-

tween political in-
stability and polit-
ical freedom:
World - .24 - .24
Africa -.03 -.04
Latin America .21 .14
Asia -.17 -.17

NOTE.-The numbe
growth rate of real
population, respecti
of government tran
DPRI3, is 1 in case t
is 3 or higher and i
cutoff point.

The IIR estimation results are reported in the second part of this
section.
Row la of table 2 presents the estimation results of equation (1)
for our total sample of 97 countries, using TRNS as a measure of
political instability. The standard errors for the coefficients are based
on H. White's heteroscedasticity-consistent covariance matrix.15 Most
coefficients are significant at the 10% level or more. The coefficient of
the political instability variable is negative, but not very significantly
so. Excluding investment increases its significance only slightly (row
I b).
We also estimated the model for three different continents: Africa,
Latin America, and Asia. In Africa political instability appears as a
very important factor in determining economic growth (rows 2a and
2b of table 2). Fosu also found for his sample of 31 sub-Saharan African

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
344 Economic Development and Cultural Change

TABLE 2

REGRESSION RESULTS: POLITICAL INSTABILITY AND ECONOMIC GROWTH, 1963-88

R2
Regression Constant K L TRNS (Adjusted) SEE
All countries:
(la) 2.13* .07* .45** -.55 .09 1.79
(.88) (.03) (.24) (.36)
(lb) 3.92* . . . 18 -.65 .03 1.85
(.48) . . . (.17) (.39)
Africa:
(2a) .77 .08 .90* - 1.33* .23 1.96
(1.45) (.05) (.44) (.52)
(2b) 1.09 . . . 1.20* - 1.77* .18 2.02
(1.49) . . . (.58) (.64)
Latin Ameria:
(3a) 1.30 .03 .93** -.34 .10 1.53
(1.60) (.06) (.50) (.61)
(3b) 1.85 . . . .92** -.44 .14 1.50
(1.33) . . . (.49) (.63)
Asia:
(4a) 5.56 .06 .13 -1.72 -.04 3.73
(3.93) (.12) (1.00) (2.14)
(4b) 7.07* . . . .14 -2.22 - .01 3.68
(2.58) . . . (1.00) (1.75)

NOTE.-The model estimated is eq. (1), where our

(TRNS) is based on the number of government transfer
of government transfers exceeds 7 and is 0 otherwise.
other variables used. Standard errors of the regressi
* Significant at the .05 level (one-tailed).
** Significant at the .10 level (one-tailed).

countries over the period 1960-86 that po

hampered economic growth. However, if al
ical instability are used, the results change
using TRNS6-this variable is assigned a valu
government transfers exceeds six (the aver
transfers in Africa) and zero otherwise-y

Q = 0.42 + 0.09K + 0.91L - 0.51 TRNS6

(1.50) (0.05) (0.48) (0.50)
(2)
R2 = 0.37 SEE = 1.28.

In Latin America the coefficient of TRNS is not significant (row 3a of

table 2). This result hardly changes when investment is excluded (row
3b). In Asia the coefficient of our political instability measure is also
not significantly different from zero (row 4a).16 Note, however, that
when investment is excluded, the significance of the coefficient for
TRNS increases, suggesting that political instability may after all affec

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
Jakob de Haan and Clemens L. J. Siermann 345

growth through its impact on the investment-in

ings are not altered if other measures of politic
In conclusion, we find some support for the
instability directly hampers economic growth in Af
America and Asia. Our results, therefore, sugge
relationship between economic performance and
that is generally measured at a global level m
smaller geographic areas. Furthermore, the resu
respect to the definition of the political instabili
Kormendi and Meguire have found a strong
the level of the investment-income ratio and their
variable. We take up this line of research by est
of the IIR. Following Kormendi and Meguire, we
the IIR, in which our proxy for political instabili
tory variable. Table 3 presents the estimation r
are consistent with those reported in table 2. It
that in Africa and Asia political instability is ne
IIR. These conclusions are not sensitive with resp
of the political instability measure. Using altern
the definition of the political stability measure
conclusions. Our findings imply that in Africa an
bility may affect economic growth in indirect w

IV. Political Freedom and Economic Growth

The first row of table 4 presents the estimation results for our tota
sample of 97 countries, using DPRI3 to differentiate countries on th
basis of political freedom.17 Although the coefficient of DPRI3 is neg
tive, it is not significantly different from zero. Kormendi and Megui
found in their regression for a sample of 47 countries a (marginally
significant) negative coefficient of Gastil's civil liberty index. How

TABLE 3

REGRESSION RESULTS: POLITICAL INSTABILITY AND THE IIR, 1963-88

Regression Constant TRNS R2 (Adjusted) SEE

(1) All countries 18.30* .00 -.01 8.87
(1.31) (1.80)
(2) Africa 14.85* - 5.99* .09 7.67
(1.57) (2.17)
(3) Latin America 18.38* -3.47 .06 5.46
(1.48) (2.26)
(4) Asia 25.32* -8.82* .17 8.88
(3.65) (4.13)

NOTE.-The model estimated is IIR = a, + a2 TRNS + e.

Standard errors of the regression (SEE) are shown in parentheses.
* Significant at the .05 level (one-tailed).

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
346 Economic Development and Cultural Change

TABLE 4

REGRESSION RESULTS: POLITICAL FREEDOM AND ECONOMIC GROWTH, 1963-88

R2
Regression Constant K L DPRI3 (Adjusted) SEE
All countries:
(la) 1.77* .06* .64* - .55 .09 1.79
(.87) (.03) (.28) (.54)
(Ib) 3.43* .46 -.77 .03 1.85
(.34) (.26) (.56)
Africa:
(2a)1.19 .10* .91* -1.40 .21 1.98
(1.32) (.05) (.47) (1.13)
(2b) 1.31 1.35* -1.39 .09 2.13
(1.38) (.58) (1.29)
Latin Ameria:
(3a)1.94 .02 .88** -.92 .18 1.46
(1.78) (.06) (.46) (.71)
(3b) 2.27 .87** - .96 .22 1.43
(1.28) (.44) (.61)
Asia:
(4a) 3.97 .08 - .41 2.68** .02 3.63
(3.39) (.11) (.99) (1.37)
(4b) 5.58* - .48 2.99* .03 3.61
(2.24) (1.00) (1.27)

NOTE.-The model estimated is eq. (1), usin

dom. Standard errors of the regression (SEE)
* Significant at the .05 level (one-tailed).
** Significant at the .10 level (one-tailed).

ever, their results were highly depen

vestment-income ratio, suggesting th
growth operates through the investm
ment is left out, both the coefficien
increase, although the coefficient rem
suggests that political repression may
its influence on capital formation. Be
dence that supports this hypothesis.
Rows 2-4 of table 4 show the resu
and Asia, respectively. It follows tha
for our total sample, the coefficient
and significantly different from zer
investment is excluded, the magn
hardly change. Note, however, that
is positive and significantly different f
Table 5 reports the estimated IIR m
and Meguire, the only explanatory v
The results are quite consistent with
total sample is there evidence that th
proxy for political freedom.

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
Jakob de Haan and Clemens L. J. Siermann 347

TABLE 5

REGRESSION RESULTS: POLITICAL FREEDOM AND THE IIR, 1963-88

Regression Constant DPRI3 R2 (Adjusted) SEE

(1) All countries 21.97* -6.85* .14 8.17

(1.07) (1.64)
(2) Africa 12.87* .39 -.03 8.14
(2.91) (3.24)
(3) Latin America 18.13* -2.75 .02 5.57
(.97) (2.15)
(4) Asia 18.64* 1.88 -.04 9.92
(3.45) (4.30)

NoTE.-The model estimated is IIR = aI + a2 DPRI3 +

Standard errors of the regression (SEE) are shown in parenth
* Significant at the .05 level (one-tailed).

To examine whether our findings are sensitive with

the construction of our political freedom dummy, we
equations in tables 4 and 5, using a dummy (DPRI4) bas
Gastil's index) istead of three as the cutoff point. This g
the same outcomes (not shown). However, for Latin Am
sults changed dramatically. The growth equation becom

Q = 3.45 - 0.03K + 0.64L - 1.76 DPR14

(1.48) (0.03) (0.42) (0.60)

R = 0.37 SEE = 1.28. (3)

Finally, we reestimated all equations in tables 4 and 5 and inclu

the measures for instability and civil liberties simultaneously. Ba
found that in that case civil liberties lose their significance in the grow
regression. For our total sample of countries our equations are

Q = 2.38 + 0.06K + 0.60L - 0.66 TRNS - 0.69 DPR13

(0.85) (0.03) (0.27) (0.36) (0.53)
(4)
R2= 0.11 SEE= 1.77

and

IIR = 23.09 - 1.74 TRNS - 7.27* DPR13

(1.49) (1.62) (1.58)(5)

R2=0.14 SEE=8.17.

The coefficient and significance of DPRI3 are very sim

reported in tables 4 and 5, respectively. Indeed, in the

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
348 Economic Development and Cultural Change

IIR equations for the other country groups, the significance o

remained the same as reported in tables 4 and 5 (not show
available from us on request). Thus, simultaneously including
stability and political freedom measures does not change our
sions.
In conclusion, we find little support for the view that po
repression negatively affects economic growth in our regiona
Only in Latin America is there some evidence that political re
hampers economic growth, but this conclusion is very sensit
respect to the construction of the political repression variable
political repression is positively associated with economic
Again, there are marked differences in the results for the entire
of countries and those for various smaller geographic areas.

V. Summary and Suggestions for Future Research

This article has three purposes: first, to examine whether the e
relationship between political instability and political freedom
nomic growth, generally measured with global data, also hold
smaller geographic areas; second, to analyze whether estimat
sults are sensitive with respect to the delineation of measures for
cal instability and freedom; and third, to investigate the relat
between political stability and political freedom and capital fo
in order to determine whether the political-institutional variable a
economic development through the investment ratio.
Using data for a sample of 97 countries for the period 1963
examined whether lack of political stability and lack of politic
dom are negatively correlated with economic growth. We estim
cross-section model based on a simple neoclassical production
tion. Our measures of political instability are based on the total
of government changes. Two dummies based on Gastil's politica
index are used to measure political freedom. A first conclusion
results for various regional groups of countries are often very dif
We find that only in Africa is there mixed support for the v
political instability reduces economic growth, both direct
through its effect on capital growth. However, the results in the
equation for Africa are rather sensitive with respect to the defini
the political instability measure. In Asia there is evidence that
instability hampers investment. There is also some mixed ev
that political repression reduces economic growth in Latin A
this result, however, is highly dependent on the constructio
political repression variable. Political repression and growth a
tively associated in Asia.
An important issue that is not addressed in this article is ca
Following recent literature, we have based our estimates on t
sumption that political instability and repression may hampe

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
Jakob de Haan and Clemens L. J. Siermann 349

nomic growth. However, the causality may be

that is, the record of economic growth may af
Mancur Olson, for instance, argues that

Rapid economic growth, whatever the nature of

must involve fast and deep changes in the ways t
the places that things are done, and in the dist
prestige. Most people spend such a large proporti
ing for a living and draw such a large part of t
political influence from their economic position t
nomic order must have great effects on other f
fore, until further research is done, the presump
economic growth, far from being the source of do
it is sometimes supposed to be, is rather a disrup
force that leads to political instability.8

It is clear that the results reported here do not

since if there is any significant correlation at
growth rates and political instability. However
disappointing growth performance leads to poli
ture research this issue should be addressed.
In this article we have used a very simple growth model to exam-
ine the impact of political instability and lack of freedom on economic
growth. Further research is needed in which more elaborate growth
and investment models are used to examine the robustness of our
findings.

Notes
* We would like to thank C. A. de Kam and the editor and two referees
of this journal for their helpful comments on a previous version of this article.
1. David Marquand, "Political Institutions and Economic Performance,"
in Government and Economies in the Postwar World, ed. Andrew Graham
and Anthony Seldon (London: Routledge, 1990).
2. Larry Sirowy and Alex Inkeles, "The Effects of Democracy on Eco-
nomic Growth and Inequality: A Review," Studies in Comparative Interna-
tional Development 25, no. 1 (Spring 1990): 126-57.
3. Gerald W. Scully, "The Institutional Framework and Economic Devel-
opment," Journal of Political Economy 96 (October 1988): 652-62; Roger C.
Kormendi and Philip G. Meguire, "Macroeconomic Determinants of Growth:
Cross-Country Evidence," Journal of Monetary Economics 16, no. 2 (Septem-
ber 1985): 141-63; Kevin B. Grier and Gordon Tullock, "An Empirical Analy-
sis of Cross-National Economic Growth, 1951-80," Journal of Monetary Eco-
nomics 24 (August 1989): 259-76; and Robert J. Barro, "A Cross-Country
Study of Growth, Saving and Government," Working Paper no. 2855 (National
Bureau of Economic Research, Cambridge, Mass., February 1989).
4. Raymond D. Gastil, Freedom in the World (Oxford: Clio), various vols.
See Sec. II for a discussion of these data.
5. Barro, pp. 21-22.
6. Augustin Kwasi Fosu, "Political Instability and Economic Growth:

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms
350 Economic Development and Cultural Change

Evidence from Sub-Saharan Africa," Economic Development an

Change 40 (July 1992): 829-41.
7. This model is very similar to the one used by Fosu, who also
exports as an explanatory variable. However, the inclusion of expo
aggregate production function may be questioned. We are not awa
sound theoretical justification for this approach and therefore ha
cluded exports in our model. As suggested by one of the referees,
also estimated the model by using the growth rate of investment m
growth rate of the population, to take account of the correlation of
hand-side variables. This did not, however, affect our general concl
8. International Monetary Fund, International Financial Statist
book (Washington, D.C.: IMF, 1990 and 1991), vols. 43 and 44.
9. R. Summers and A. Heston, "The Penn World Table (Mark 5): An
Expanded Set of International Comparisons, 1950-1988," Quarterly Journal
of Economics 106 (May 1991): 327-68.
10. Sebastian Edwards and Guido Tabellini, "Explaining Fiscal Policies
and Inflation in Developing Countries," Journal of International Money and
Finance 10, suppl. (1991): S16-S48. We have also constructed a similar dichot-
omous variable based on the number of irregular government transfers (coups).
Since this variable generally yielded the same outcomes, we do not report
them here.
11. C. L. Taylor and D. Jodice, World Handbook of Social and Political
Indicators (New Haven, Conn.: Yale University Press, 1983).
12. Arthur S. Banks, Political Handbook of the World (New York:
McGraw-Hill), various issues.
13. We have also experimented using an average of five as a cutoff point,
but this yielded similar results as for the models with PRI4.
14. Kormendi and Meguire (n. 3 above), p. 46.
15. H. White, "A Heteroskedasticity-Consistent Covariance Matrix Esti-
mator and a Direct Test for Heteroskedasticity," Econometrica 48 (May 1980):
817-38.
16. Note the very poor fit of the regression for Asia. Apparently the model
is not able to capture the highly diverging development patterns in Asia. We
have tried various country-dummies to improve the fit, but this did not change
our basic conclusion with respect to the significance of the coefficient of
TRNS.
17. See Sec. II for further details on the construction of this dummy.
18. Mancur Olson, "Rapid Economic Growth as a Destabilizing Force,"
Journal of Economic History 23 (1963): 519-52, quote on 550.

This content downloaded from 213.55.95.148 on Fri, 25 Jan 2019 08:58:07 UTC
All use subject to https://about.jstor.org/terms