You are on page 1of 14

Economic Systems 34 (2010) 91104

Contents lists available at ScienceDirect

Economic Systems
jo urnal homepage: www.elsevier.com/loc ate/ecos ys

Capitalism and freedom?


Frederic L. Pryor *
Swarthmore College, Economics, 500 College Avenue, Swarthmore, PA 19081, United States

A R T I C L E I N F O

A B S T R A C T

Article history:
Received 6 March 2009
Received in revised form 11 September 2009
Accepted 19 September 2009

This essay tests Milton Friedmans conjecture that capitalism is a


necessary condition for political freedom. For the decade around
2000 indices of the degree of capitalism and the degree of political
freedom are highly correlated and provide plausibility for Friedmans conjecture. In looking at changes over time in the nineteenth
century, however, the analysis refutes Friedmans conjecture. These
apparently contradictory results are reconciled by showing that
both capitalism and freedom are related to such variables as the
educational level of the population so that, although not causally
tied, they are correlated in a cross-national comparison.
2009 Elsevier B.V. All rights reserved.

JEL classication:
P10
P16
P51
N10

Keywords:
Capitalism
Economic freedom
Political freedom

1. Introduction
Over the last two centuries, as we all know, capitalism has spread throughout the world and, at the
same time, the level of political freedom has risen as well, albeit unsteadily. The relationship between
these two trends has been, for the most part, a relatively closed topic for serious analysis1 since
Friedman (1962, pp. 910) wrote his oft-cited words when analyzing the intimate connection
between economics and politics and the fact that only certain combinations of political and
economic arrangements are possible: Economic freedom is also an indispensable means toward the
achievement of political freedom. . . Historical evidence speaks with a single voice on the relation
between political freedom and a free market. I know of no example in time or place of a society that

* Tel.: +1 610 328 8130; fax: +1 610 328 7352.


E-mail address: fpryor1@swarthmore.edu.
1
A number of analyses have, however, examined the relationship between income and freedom, and their results are
relevant to this study since the degree of capitalism and the level of economic development are related.
0939-3625/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecosys.2009.09.003

92

F.L. Pryor / Economic Systems 34 (2010) 91104

has been marked by a large measure of political freedom, and that has not also used something
comparable to a free market to organize the bulk of its economic activities. . . History suggests only
that capitalism is a necessary condition for political freedom. Clearly it is not a sufcient condition.
These ideas, also argued by others (e.g., Hayak, 1944, pp. 6970), raise a number of theoretical and
empirical difculties.
Certain theoretical problems arise if we consider the possibility of joint causation, a likelihood
Friedman (1991) raises in his Smith Center lecture that while economic freedom facilitates political
freedom, political freedom, once-established, has a tendency to destroy economic freedom.2
Moreover, it is unclear about the time-period that must be taken into account to understand these
ideas.
Other empirical problems in examining these ideas also arise. Capitalism and freedom are
correlated with a number of common variables, so that separating their joint impact is difcult.
Country xed effects also muddy the relationship. Finally, although Friedman speaks of historical
evidence, he does not actually supply any. Difculties arise in knowing what he means by a large
measure of political freedom and quantifying either capitalism (which he equates with economic
freedom) or political freedom in a way to allow empirical investigation. Indeed, obtaining the proper
data for such an analysis is probably the most important reason why the relationship between
capitalism and freedom has been neglected in the literature. Ideally we would like a large panel study
of nations reaching back several centuries and offering data for both concepts, but, unfortunately, such
an extensive data set does not currently exist.
This essay attempts to pin down the relationship between capitalism and freedom and to extract
some stylized facts to serve as a more solid foundation for analyses of the relationships between these
two variables, which are so critical for our understanding of how the economic system works. It starts,
therefore, with a clarication of capitalism and political freedom and shows how they can be
quantied. For a cross-section comparison in recent years, I use a relatively complete measure for both
concepts; for the time-series analysis using nineteenth century data, less satisfactory measures must
be used for both and are described later in the essay.
For the cross-country comparisons I search for exceptions to Friedmans conjecture in a
straightforward manner and show that capitalism and freedom are positively and signicantly related
in 2000, with only few notable exceptions. Then I turn to the time-series data for the nineteenth
century and nd that such a relationship does not occur. This result casts grave doubt on any causal
connection between capitalism and freedom and suggests that the cross-section relationship is due to
country specic variables (as is the case of the relationship between income and political freedom, as
shown by Acemoglu et al., 2008) and to the fact that both capitalism and political freedom are related
to per capita GDP, the educational level of the population, and other factors.
2. Conceptual clarications and quantication
Capitalism has a multitude of denitions, but many are difcult to quantify. Nevertheless, by using
a series of indicators, each standardized to have a mean of zero and a standard deviation of one, the
available indicators can be easily combined so that for each country we can determine the degree to
which it approaches a pure capitalist system. The same approach is also used to determine the degree
of political freedom. Thus, both cross-section and time-series comparisons can be made between the
two variables for recent years.
Since capitalism is a structure, it is important in calculating the index to avoid selecting elements
that reect, for instance, particular policy instruments or indicators of economic success, such as
stable prices and exchange rates, economic efciency, or equity. This is because many of these policies
and results appear in both capitalism and other types of economic systems. In some cases,
unfortunately, such as determining the amount or degree of governmental economic intervention in
2
This claim, which seems to muddy his earlier argument, was made in the context of his comparison between Hong Kong
(which, he claims, had economic, but not political freedom) with India (which, he points out, had political freedom but not
economic freedom). The Indian example also contradicts his thesis that economic freedom is indispensable for political
freedom.

F.L. Pryor / Economic Systems 34 (2010) 91104

93

the nineteenth century, such an approach proved impossible because of lack of comparable data for
many countries with which to construct appropriate indicators. As a result of these data problems, the
denitions of capitalism in the year 2000 and in the nineteenth century are somewhat different.
Nevertheless, the general conclusions are comparable.
2.1. Elements of modern capitalism
2.1.1. Minimum level of per capita production
An important prerequisite for extensive trade is production using capital-intensive methods that
permit a considerable surplus of production over the food and other basic goods necessary for life.
Although highly commercialized agricultural societies might conceivably achieve such a surplus, in
the vast majority of cases a signicant volume of tradable goods implies a level of per capita
production and industrialization not possible with handicraft technology. Difculties arise in trying to
specify the minimum level of per capita production needed for such a tradable surplus, especially
because of problems in determining the total production of goods and services in countries at low
levels of economic development where a large part of production is for home consumption. This
means that only a part of the total production of goods and services reaches the market place so it can
be easily measured; the remainder must be estimated (or neglected) by national account
statisticians.3
My estimation of the minimum measured per capita GDP necessary to allow extensive trade is
based on two sources: Maddisons (2003) estimates of per capita GDP in 1990 GearyKhamis dollars
for over 160 political entities for a period spanning in some cases over 500 years; and Adelman and
Morris (1978) estimates for many countries in the nineteenth century of the degree of to which
exchange of goods, labor, land, and nance pass through markets. Transforming the Adelman and
Morris data to a scale running from 0 through 10 and extrapolating to 1816, Great Britain had an
average marketization score for the four markets of ve.4 Most economic historians would agree that
the geographic region now designated as Great Britain certainly had a capitalist economic system in
1816, and in that year it had a per capita GDP of $1689. I choose this income level as my criterion for
when a country passes the threshold so that the other three criteria for capitalism can be applied to
determine the degree to which the country has a capitalist economic system.5 In brief, this minimum
level of per capita GDP serves to screen out those countries that are not capitalist, but otherwise does
not play a role in determining the degree to which the country has a capitalist economic system.
2.1.2. Substantial private and legally protected ownership of the means of production
Friedman argues that private ownership of the means of production is essential for economic
freedom, and two aspects of this element of capitalism deserve to be noted. First, to qualify as a
capitalist economic system, ownership of most of the means of production must ultimately be in the
hands of private individuals, acting singly or in conscious groups (e.g., corporations), rather than in the
hands of some all-embracing organization such as the government or a church. Ownership may be
separated from control of such productive capital, but the private owners have the ultimate decisionmaking authority. If the productive capital is used by hired labor, this means that society is divided
into at least two groups, the hired labor and the owners of the means of production, who receive the
bulk of the prots arising from the use of such capital. Second, both this ownership and the prots
3
Although such home production is relatively more important in low-income countries, even in the United States in the
1960s, time-budget data from Sirageldin (1969, p. 55) show that the GDP was understated by very roughly 16% because of the
exclusion of home production/consumption.
4
The AdelmanMorris estimates are derived by extrapolating backward for the base years of 1850 and 1870. Estimating the
degree of commercialization is, of course, a difcult task and for England in the eleventh century there is considerable
controversy, as seen in the debate between various authors in Britnell and Campbell (1995). The AdelmanMorris estimates
have the advantage of using consistent criteria for all countries, even though some of their evaluations may be questioned. The
GDP estimate discussed in the next paragraph is derived by extrapolating backwards from Maddisons (2003, p. 59) estimates
for 1820 and 1830.
5
This approach means that Englands industrial revolution started before the country would be considered capitalist by this
criterion, because market exchange did not account for a majority of the countrys exchange of goods, services, and factors of
production at that time.

F.L. Pryor / Economic Systems 34 (2010) 91104

94

derived are protected by laws, which are effectively enforced by the government. Although the legal
protection of property is also an aspect of political freedom, in dening political freedom I do not
include such legal protection.
For legal protection I use six indicators: a rating of the law and order in a country, its rule of law,
the independence of its judiciary, the impartiality of its courts, its protection of intellectual property,
and its legal enforcement of contracts. Sources of these indicators and the way I have handled them, as
well as for all other criteria discussed below, are presented in Appendix 1. These six indicators for legal
protection of property are highly correlated and are shown in Part A of Appendix Table A1 on my
website (Pryor, 2009). I give each of them equal weight in a property protection index (Gwartney et al.,
referenced 2007).
Comparable data on the extent of public and private ownership of the means of production for the
countries in my sample are not available. As a proxy, I use private investment as a share of total
investment, a series which is only weakly (but signicantly) correlated with the legal protection index.
These two indices are combined, with equal weights, to calculate the index for legally protected and
substantial private ownership of the means of production.
2.1.3. Competitive markets
Economic freedom and, hence, capitalism require a competitive-market system with uctuating
prices serving to allocate a substantial share of the goods and services produced in the economy, as well
as a large share of the factors of production (labor, land, and capital). Competitive means that buyers
and sellers in the markets for factors of production and for goods and services must have free entry into,
and out of, the market and must be able to compete against each other. This implies that an economy of
monopolies or oligopolies is not capitalistic, nor is one which sets up signicant barriers to entry or exit.
My four indicators of competitive markets are: the extent of price controls, competition in
domestic banking, whether collective bargaining is carried out at the central or local level, and a
business freedom index containing various measures of the costs of starting a business, obtaining a
license, and closing a business. The data come from studies of the Fraser Institute and the Heritage
Foundation (Heritage Foundation, referenced 2008; Beach and Kane, referenced 2008). Unfortunately,
a measure of monopoly that is comparable for a large number of nations is not available.6 The
relationships of these four indicators of competitive markets are shown in Part B of Appendix Table A1
(Pryor, 2009).
2.1.4. Direct economic freedom
This concerns the degree to which the government interferes in the workings of the economy. Some
government functions do not count as interference: protection of property and person by the army,
police, and courts; the provision of infrastructure such as roads; anti-trust enforcement;
implementing product safety and truth in advertising regulations; providing free public education
and other goods and services with important externalities; and so forth. But when the government
regulates prices or quantities, limits entry into occupations by licenses, enforces social security
payments, redistributes income, drafts soldiers into the armed forces, such measures are deemed in
this analysis to move the economic system away from capitalism.
My measure of economic freedom includes ve indicators: trade freedom, investment freedom,
nancial freedom, labor freedom, and lack of interest rate regulation. The data come from the Fraser
Institute and the Heritage Foundation. Their interrelations are shown in Part C of Appendix Table A1
(Pryor, 2009).
2.2. Modern political freedom
Political freedom has two major elements: (a) an elected government responsive to the people (i.e.,
democracy); and (b) civil liberties. The former refers to the manner in which the head of the
government and the legislature are chosen, the determination of who can participate in this process,
the ease with which people can organize political parties, the constraints that are placed on executive
6

Such comparable data are available for European countries for earlier years (e.g., Pryor, 1972), but for few others.

F.L. Pryor / Economic Systems 34 (2010) 91104

95

actions, and the extent to which governmental actions are transparent, accountable to the electorate
and free of interference from non-elected groups or individuals. Civil liberties refers to freedom of
expression and belief, associational rights such as freedom of assembly, the rule of law in civil and
criminal proceedings, and personal autonomy, for instance, the right to travel or live as one chooses.7
In assessing political freedom, it is necessary to focus on what is actually happening, rather than
what is written in the constitution or laws. For political freedom in the modern era, four indicators of
political freedom incorporating in different proportions the above-discussed conditions are available
for us to draw upon. As shown in the correlation matrix in Part D of Table A1 on my website, all of the
indicators are highly correlated with each other for the nations in the sample.
The rst indicator comes from the Polity IV project, estimated by the Center for International
Development and Conict Management of the University of Maryland (2008) and directed by Monty
G. Marshall and Keith Jaggers. Their Polity2 measure focuses primarily on democracy (political
freedom) and extends back to 1800 for independent nations, that is, countries that are not under
colonial domination. Their scale is based on separate ratings of the competitiveness of executive
recruitment, openness of executive recruitment, the constraint on the chief executive, restriction of
the participation in election, and competitiveness of political participation. Their sample includes
almost 190 independent political entities (not colonies) and, depending on how long they have been
independent, from 1800 onward and for the years from 1995 to 2004 (their last year) almost all
nations in the capitalist sample were independent.
The second and third indicators were rst calculated by Raymond D. Gastil and the results
published by Freedom House (2008). These are separate indices for responsive government (which
they call political rights) and civil rights and run from 1970 to the present. I have reversed the indices
so that a high number indicates high-political freedom, as is the case with the other indices in my
analysis. The nal indicator is voice in, and accountability by, government and has been calculated
by Kaufmann et al. (2007) for the World Banks governance project. This is a composite indicator that
includes some of the other three series plus a number of other indicators as well.
As in the case of the indicators for capitalism, each of the four indicators was standardized so that the
mean for the sample nations is zero and the standard deviation is one. These four transformed indicators
were given equal weights and then averaged, since we have no measure of the relative accuracy of each.
Then this averaged series was also standardized. Such a procedure minimizes any special biases of any of
the four indicators. The nal results are presented in my web page (Pryor, 2009).
2.3. Sample selection
Although World Bank statistics list over 200 political entities, I chose slightly less than half of them
for this study. My four criteria of selection were: (a) independence; that is, the nation had to be
sovereign, not a colony; (b) population of at least one million, so that tiny countries with highly
idiosyncratic characteristics would be eliminated; (c) available data for the indicators used in this
study; (d) a minimum level of per capita GDP, as previously noted. As a check on Maddisons estimates
for 19952005, I also used World Bank (2008) estimates of per capita GDP in international dollars
(with suitable adjustments so that the two series were commensurable).

3. A cross-section of political freedom and capitalism in the decade around 2000


To gain an idea about the relationship between capitalism and political freedom, I begin with a
simple cross-section regression and then look for notable exceptions. Although such a crude approach
tells us little about causal relations, it sets up the questions to be answered using an analysis over time
in the next section.
7
Rothschild (2003) emphasizes that Friedmans discussion of freedom focuses primarily on negative freedom, i.e., freedom
from constraints on an individuals ability to make the most of his capacities and opportunities according to his own lights,
subject to the proviso that he not interfere with other individuals to do the same. Friedman does not consider positive
freedom, which concerns opportunities. Rather he focuses on constraints; similarly, the indicators of political freedom used in
this essay also focus primarily on negative freedom.

F.L. Pryor / Economic Systems 34 (2010) 91104

96

Graph 1. Political freedom and capitalism. Notes: Both political freedom and the degree of capitalism variables are transformed
so as to have a mean of zero and a standard deviation of one. Countries having high-political freedom and low-capitalism scores
(i.e., exceptions to the Friedman conjecture) are in the southeast quadrant. Countries with high capitalism but low political
freedom are in the northwest quadrant. To maintain legibility in this graph, slightly less than 40% of the points are labeled.

3.1. Cross-section relationships


The scatter-diagram of political freedom and capitalism in Graph 1 shows a roughly linear
relationship. These data can be used to set up a baseline relationship between these two variables for
beginning the analysis.
Tables 1 and 2 present OLS regressions linking modern capitalism to political freedom in 2000 and,
for the sample as a whole, reveal a positive and statistically signicant correlation. Robust regressions
and weighted regressions (both by population and the log of the population) yield quite similar
coefcients for the capitalism and per capita GDP variables. In Part A of Table 2 the sample is
Table 1
Political freedom and capitalism around 2000.
Constant

Capitalism

Dependent variable = political freedom


(1)
0.000 (0.082)
0.636* (0.082)
(2)
0.000 (0.080)
0.453* (0.111)

Per capita GDP

Sample size

Adj. R2

0.265* (0.111)

90
90

0.3975
0.4281

Notes: In these OLS regressions the capitalism variables and the per capita GDP variables are transformed so that their respective
means are zero and the standard deviation is one. For this reason, the calculated coefcient for the constant for the full sample is
zero and the two standard errors of the independent variables are equal.
The per capita GDP data come from Maddison (2003) and are an average for the period 1990 through 1994. The sources of other
data are described in Appendices.
*
Denotes statistical signicance at the 0.05 level.

F.L. Pryor / Economic Systems 34 (2010) 91104

97

Table 2
Political freedom, capitalism, and economic freedom around 2000.
Sample

Per capita
GDP range

Part A: Degree of political freedom = dependent variable


(1)
Full sample

Constant

Economic
freedom

Sample
size

Adj. R2

0.000
(0.082)

0.636*
(0.082)

90

0.3975

>$1,689
<$4,500

0.225
(0.153

0.652*
(0.178)

33

0.2795

(2)

Low income

(3)

Middle income

>$4,500
<$13,000

0.106
(0.140)

0.564*
(0.167)

38

0.2194

(4)

High income

>$13,000

1.107*
(0.235)

0.062
(0.169)

19

0.0505

Sample

Per capita
GDP range

Constant

Political
freedom

Part B: Economic freedom (dependent variable) and political freedom in 2000


(5)
Full sample
All
0.000
0.662*
(0.079)
(0.080)

Sample
size

Adj. R2

90

0.4322

(6)

Low income

>$1,689
<$4,500

0.274
(0.148)

0.624*
(0.168)

33

0.2844

(7)

Middle income

>$4,500
<$13,000

0.128
(0.130)

0.627*
(0.143)

38

0.3303

(8)

High income

>$13,000

1.194*
(0.250)

0.217
(0.192)

19

0.0154

Notes: The data and their sources are described in Appendices. Both the economic and political freedom variables are
transformed so their means for the entire sample are respectively zero and their standard deviations and one. Because this
standardization is not repeated for the subsamples, the calculated coefcients for the constant are not zero. The standard errors
are given in parentheses and statistical signicance at the .05 level is designated with an asterisk.

subdivided into income classes and this relationship also holds for low and middle income nations, but
not for high income nations, where the degree of capitalism and the political freedom scores are
roughly the same for all countries.
Many cross-section regressions by others show that per capita GDP is signicantly related to political
freedom.8 The second regressions in Table 1 indicate that after adding per capita GDP, the signicantly
positive relationship between political freedom and capitalism persists. Of course, the single-equation
models presented in Table 1 do not show the direction of causality, and, as I later argue, none exists in
either direction; the relationship between capitalism and political freedom appears to arise because both
are related to other variables, such as education.9 Nevertheless, the relationship shown in the regressions
between political freedom and capitalism provides support (albeit misleading) for Friedmans basic
intuition. For those nations which became capitalist in the nineteenth century, I also tested whether the
degree of capitalism is related to the length of time that a nation has been capitalist. There was, however,
no signicant relationship, primarily because the nations in this subsample all had high degrees of
capitalism in 2000 and hence there was little variation to explain.
Because Friedman often denes capitalism solely in terms of economic freedom without reference
to ownership of capital or degree of competition, it is useful to examine the relationship between
economic freedom and political freedom. As Part B of Table 2 shows, the two types of freedom are
signicantly related. Again, it is unclear in which direction causality runs and, in many cases, changes
in both types of freedom are sometimes caused by the same external forces. For instance, in wartime
8
Acemoglu et al. (2008) briey discuss the literature concerning political freedom and income, and have correctly argued
that such regressions as those in Table 1 do not reveal the entire story: in fact, once country xed effects are taken into account,
the relationship between per capita income and political freedom is no longer statistically signicant.
9
Pryor (2005, Chapter 5) presents evidence showing that both industrialization and marketization are strongly related to
literacy. But so is political freedom as well.

98

F.L. Pryor / Economic Systems 34 (2010) 91104

or periods of national emergency, both political and economic freedom often decline and then
rebound together when the crisis is over.
When I ran cross-section regressions for 1850, 1870, and 1890 using somewhat different measures
for capitalism and political freedom (see below), the signicantly positive relationship between the
two variables general results also appeared. Although the calculated coefcients for the capitalism
variable differed somewhat from those shown in Table 1, they all varied within a relatively narrow
range. I also experimented with a number of different specications of the regressions discussed
above. For instance, I added a dummy variable indicating that the nation was in transition from a
centrally planned economy, but none of these experiments turned out to reveal anything of interest.
3.2. Possible exceptions to Friedmans conjecture
Since Friedman claimed that capitalism is a necessary condition for political freedom, a crude way
to nd exceptions is to look for nations which have a capitalism score that is X standard deviations
higher than the sample average while, at the same time, having a political freedom score that is X
standard deviations lower than the sample average. Setting X = 0.5, Kuwait, Singapore, and the UAE fall
into this class.10 Thus capitalism does not appear a sufcient cause of political freedom. The reverse
case is when a nation has a capitalism score that is lower than X standard deviations from the sample
average while, at the same time, having a political freedom score that is X standard deviations higher
than the sample average. With X = 0.5, only Slovenia ts this bill of having political freedom without
much capitalism, at least for the decade under examination.11
A more exact way of examining the issue is to use equation 2 in Table 1 (so as to hold per capita GDP
constant, since it has a possible impact on both variables) to derive a predicted score for political
freedom and then to determine the difference with the actual political freedom score. By dividing this
difference by the standard error of estimate (root mean square error) of the equation, we can carry out
a simple t-test using a 0.95 level of condence.
For four countries, namely Kuwait, Oman, Singapore, and the UAE, the predicted political freedom
score is signicantly less than predicted, and for no country is the predicted political freedom score
signicantly more than predicted. The rst result provides evidence for Friedmans assertion that
capitalism is a necessary but not a sufcient condition for political freedom. The second nding
suggests that capitalism appears a necessary condition or accompanying factor for political freedom.
In the cross-section regressions for 1850, 1870, and 1890, there are also very few exceptions to
Friedmans conjecture that were statistically signicant; moreover, they seemed to reect temporary
conditions since these low-capitalism, high-political freedom countries were different in each of the
three years for which the regressions were run. I also carried out a similar exercise, but looking at the
relationship between economic and political freedom. Finally, I also carried out the same exercise
(including holding per capita GDP constant), but looked instead at the relationship between political
and economic freedom in 2000. The results were little different from those using capitalism as the
independent variable.
In sum, the cross-country evidence from single equation OLS regressions ostensibly provides
support for Friedmans claims. But such an approach does not take into account that causation may
run in the opposite direction or that the degree of capitalism and political freedom is related to some
variable that is the same for both.
4. Political freedom and the emergence of early capitalism: an analysis over time
For the time-series analysis, the lack of suitable data requires a redenition of both capitalism and
political freedom somewhat different than before. Then, I turn to Friedmans argument, examine
briey some issues about the origins of capitalism, and nally I deal with the apparent inconsistency
between the time-series and cross-section results.
10

Kuwait, Singapore, and the UAE have Polity2 scores, respectively of 7, 2, and 8 for the decade under consideration.
Friedmans argument that capitalism is a necessary condition for political freedom is not really refuted by the Slovenian
case because this particular situation might be attributed primarily to a slow transition from its previous communist economy.
11

F.L. Pryor / Economic Systems 34 (2010) 91104

99

4.1. Early capitalism


For the nineteenth century, comparable cross-national data on capitalist institutions are usually
not available. For instance, we have no measures of economic freedom that could be used for all of the
countries in the sample (although for a much smaller sample, certain measures, such as tariff rates, are
available for many countries). According to Chang (2005, p. 280), Virtually all of todays developed
countries actively used interventionist industrial, trade, and technology policies aimed at promoting
their new (infant) industries against the competition from the established industries based in more
advanced nations. But there is no easy way to quantify such interventions. The other elements of the
denition of capitalism provide fewer problems. The degree to which property was protected in the
nations which passed the threshold test for a capitalist economy appears very roughly similar. As for
the competitive-market element, public ownership of the means of production was relatively low in
almost all countriesmassive government ownership of capital was a twentieth century
phenomenon.
The key criterion for determining the degree of capitalism in these nineteenth century economies
which had passed the threshold limit is the average marketization score for goods, labor, land, and
nance. Adelman and Morris (1978) provide such estimates for 1850 through 1890, and by
interpolation and extrapolation estimates for this study could be made for 23 more countries in the
same manner as that used for Great Britain.12 These ratings of market penetration are correlated with
Maddisons per capita GDP estimates. Therefore, we can extend the 17 nations in the AdelmanMorris
sample of countries that had become capitalist in the nineteenth century by using the results of a
regression linking their sample with per capita GDP. This adds seven more nations to some of the
samples used below in Table 4.
4.2. Early political freedom
The only available indicator of political freedom extending back to 1800 is the Polity2 variable in
the Polity IV data base, and this indicator is designated as early political freedom. As noted above,
this variable measures the degree of democracy and, although democracy and freedom are somewhat
different, Part D of Appendix Table A1 (Pryor, 2009) shows that the Polity2 variable is highly correlated
with the overall measure of political freedom for the year 2000. Since the same causal links between
the various measures of political freedom appear to have operated in the past, for this reason the
democracy variable serves as a useful proxy for political freedom in the nineteenth century. This
indicator is coded on a scale running from 10 (autocracy) to +10 (full democracy); I do not, however,
use the standardization transformation employed in the previous discussion.
4.3. Comparisons of capitalism and political freedom
It is useful to start by focusing on the 33 countries that cannot be considered capitalist in the
nineteenth century because they do not pass the minimum per capita GDP criterion. The next step is to
search for those which have a high degree of political freedom. Some of these countries did have
positive political freedom scores in some years, but in many cases these periods of freedom did not last
long. I have, therefore, only selected those countries with positive political freedom scores for at least
15 consecutive years to avoid focusing on various short-run phenomenons.
Of the 33 countries in the non-capitalist subsample, nine nations meet this criterion of political
freedom and are listed in Table 3. For instance, Costa Rica had political freedom ratings of +1 from 1841
through 1852, of +3 from 1853 through 1866, of +5 from 1857 through 1889, and of +10 from 1890
through the end of the century. Ethiopia had a political freedom rating of +4 from 1866 through 1899.
Yet neither country had a capitalist economic system in those years because of their low level of
economic development. Two of the countries listed in Table 3 lapsed into negative political freedom
scores, namely Guatemala by the mid 1890s and Liberia by the mid 1880s.
12
Morris and Adelman (1988) use a somewhat different method of aggregation but arrive at a nal result that is almost exactly
the same as mine. My method, however, allows analysis of the separate components.

F.L. Pryor / Economic Systems 34 (2010) 91104

100

Table 3
Future development of countries with high-political freedom and low-capitalism scores in the nineteenth century.
Marketization score,
1900

Per capita GDP,


1900

Capitalism score,
2000

Countries with 15 or more consecutive years with freedom scores in the: nineteenth century
1 or 2
Bolivia
n.a.
n.a.
0.53
Guatemala
n.a.
n.a.
0.30
Korea, South (area)
3.4
$731
0.15
Peru
3.5
817
+0.07
Serbia
3.7
902
0.86

Per capita GDP,


2000

$2,575
3,396
14,343
3,686
2,354

3 or 4
Ethiopia
Liberia
Orange Free State

n.a.
n.a.
4.7

n.a.
n.a.
1,440

n.c.
n.c.
+0.00

624
847
4,139

5 and over
Costa Rica

n.a.

n.a.

+0.05

6,174

Notes: n.a. = not available; n.c. = not capitalist.


The political freedom (democracy) scores come from Marshall and Jaggers (2005), where countries are rated from 10
(autocracy) to +10 (democracy). The marketization scores were estimated from data from Adelman and Morris (1978) and run
from 0 to 10, with 5.0 considered as borderline capitalist in this study; they are standardized for the entire 90 countries in the
sample to have a mean of 0.00 and a standard deviation of one. The per capita GDP data come from Maddison (2003) and are in
1990 international GearyKhamis dollars. The estimation of the capitalism scores is discussed in the text; a full listing is
presented in Appendix Table A2.
The per capita GDP of the area of South Korea in 1900 is interpolated. For the Orange Free State, data on per capita income and
marketization are for the entire Union of South Africa. For Serbia, the 1900 per capita GDP data are for Yugoslavia as a whole.
Ethiopia includes Eritrea.

Evidence from these non-capitalist nations casts doubt on Friedmans notion that capitalism
appears as an indispensable means for political freedom. If we reverse Friedmans hypothesized
direction of causality so that political freedom necessarily leads to capitalism, this means that the nine
countries listed in Table 3 should have been rapidly heading toward capitalism, but this is refuted by
the Ethiopian and Liberian cases.
At this point, two other implications of this discussion of Table 3 deserve brief mention: Political
freedom by itself cannot bring economic development. And nothing guarantees that high levels of
political freedom can be maintained.
The refutation of Friedmans conjecture revealed by the results in Table 3 is reinforced by
comparing changes in the cross-section ratings of marketization (capitalism) using Adelman and
Morris (1978) data for 1850, 1870, and 1890 and changes in the political freedom variable. With these
data we can look at the relationship between changes in the degree of capitalism between these 20year periods and changes in the extent of political freedom during the same years or in the previous
20- or 40-year periods.
The results of such comparisons can be quickly summarized: after looking at a number of
specications of this relationship with different lags, I found no statistically signicant relationship
between changes in the degree of capitalism and the changes in political freedom.13 In brief,
capitalism and political freedom appear to have no causal relationship with each other. In these
regression experiments, both variables are, however, positively related to per capita GDP. Such results
raise further doubts that capitalism is a necessary cause of political freedom.
4.4. Origins of capitalism
Friedman has correctly noted that capitalism can begin in countries without political freedom,
and this issue can also be usefully explored with the nineteenth century data. Table 4 presents
13
This conclusion holds when, in addition to the OLS regression, robust and weighted regressions (by population and log
population) are estimated.

Table 4
Autocracy/democracy measures for early capitalist countries and capitalist threshold year (CTY).
Country

CTY

Columns nos.

Average,
total sample

1816
1829
1832
1833
1838
1847
1855
1855
1864
1869
1875
1875
1878
1881
1882
1884
1886
1887
1889
1893
1894
1894
1897
1899

In the years from 1800


to 20 years before CTY

In the 20 years
before the CTY

In the 20 years
after the CTY

In the 20 years following


the CTY up to 1920

Signicance
tests

1
Avg.

3
Avg.

4
S.D.

5
Avg.

6
S.D.

7
Avg.

8
S.D.

2.00

0.00
0.00
0.00
1.51

8.14
8.60
6.10
1.55
4.00

0.64
1.96
1.02
1.10
0.00

2.05
0.00
0.44
1.67
4.87
0.00
1.42
0.89
0.44
0.00
2.51

4.00
2.65
1.20
9.35
4.85
4.00
1.20
1.00
8.30
7.00
9.85

0.00
3.36
1.51
1.11
1.73
0.00
2.04
0.00
1.38
0.00
0.37

4.00
4.36
3.30
9.60
3.40
4.00
1.00
1.00
7.20
5.75
9.50
4.00

0.00
0.69
0.73
0.50
6.13
0.00
0.00
0.00
3.69
1.12
2.12
0.00

5.30
5.73
9.49
1.81
4.50
10.00
0.07
3.47
0.50
2.45
9.00
10.00
1.14
6.00
3.00
10.00
10.00
4.15
1.82
1.00
0.67
4.00

2.29
2.15
0.87
3.04
5.43
0.00
2.89
2.19
5.30
5.25
0.00
0.00
0.64
0.00
0.00
0.00
0.00
1.34
0.40
0.00
0.52
0.00

e, f
f

9.00
6.00
2.80

1.00
4.00
9.25
5.55
2.40
10.00
6.02
4.50
4.25
3.00
6.00

2
S.D.

5.25

2.26

7.11

1.71

9.37
10.00
8.43
9.69

0.27
0.00
0.87
0.47

5.38
3.75
10.00
7.00

3.05
2.19
0.00
0.00

4.25
8.49
2.78
4.00
5.95

0.97
3.38
1.25
3.38
5.28

4.73

1.48

0.04

c
f
b, c, e
b
b, c, d
e
d, e, f
c, e, f
b, c
c
c, e
d
b, c
a, b, c
a, e
e
a, b

F.L. Pryor / Economic Systems 34 (2010) 91104

Great Britain
Belgium
USA
Netherlands
France
Switzerland
Germany
Austria
Sweden
Denmark
Canada
Australia
Italy
Spain
Chile
New Zealand
Norway
Hungary
Argentina
Japan
Mexico
Romania
Greece
South Africa

Autocracy/democracy measures for countries

3.29

Notes: The criteria for determining the capitalist threshold year are discussed in the text; all measurements are made relative to the CTY. The data on political freedom come from the Polity2
variable of Marshall and Jaggers (2005) and run from 10 (autocracy) to +10 (democracy). The signicance tests are t-tests of differences between the means of various columns: a = columns
1 and 3; b = columns 1 and 5; c = columns 1 and 7; d = columns 3 and 5; e = columns 3 and 7; and f = columns 5 and 7. The total sample statistics give equal weight to the data for each country.

101

102

F.L. Pryor / Economic Systems 34 (2010) 91104

data on the 24 countries which, between 1800 and 1900, can be considered as having economic
systems of early capitalism, as dened above. The data are listed according to the date when the
various countries had an average marketization level of 5.0 or over, the capitalist threshold year
(CTY). For each country I also divide the years from 1800 to 1900 into four groups: 20 years before
and after they reached the capitalist threshold (the capitalist window), the years between 1800
and the earliest year of the capitalist window; and the years from the latest year of the capitalist
window to 1900. The blanks in the table denote years for which there are no Polity IV estimates,
mostly because the nation was a colony of another country or was at war, either internal or
external.
For only 15 of the 24 countries are data available on political freedom before they reached the
capitalist window (column 3); and the governments of these countries were generally autocratic. In
this period their average degree of democracy was 4.71, which is quite low. This can be compared
with a rating of 4.01 for 33 countries that cannot be classied as capitalist in the nineteenth century
and for which political freedom data are available. In brief, capitalism seemed to begin in countries
with less, rather than more, political freedom.
It is worth noting that in the early capitalist window, as the nations became more capitalistic, the
democracy/autocracy scores generally increased (columns 3 and 5), as we might expect from
Friedmans argument. The differences in the democracy/autocracy scores were, however, slight and
statistically signicant (designated d in column 9) in only a few countries such as Denmark, Italy,
and Hungary. Finally, in the years following the last year of the capitalist window, the average
democracy/autocracy scores usually rose (columns 5 and 7), but again this change was statistically
signicant (designated by an f in column 9) for only a few countries, namely, Great Britain,
Switzerland, Italy, and Spain. These data lead to the conclusion that capitalism was usually born not in
the golden glow of political freedom, but under the cloud of autocratic governments and that it took
many years before political freedom began to increase.
4.5. Parallel changes of capitalism and political freedom at a point in time and over time
As noted above, in the nineteenth century, changes in the scores for political freedom and for
capitalism do not appear strongly related, since neither variable had a direct causal inuence on the
other. On a cross-section basis the levels of capitalism and political freedom are correlated with
some of the same variables. For instance, using the years of schooling of the population 25 and over
as a measure of human capital, we nd that such a measure accounts for 4458% of the variance of
political freedom, depending on the time-lag. This schooling variable also accounts for 4451% of
the variance of capitalism, depending on the time-lag.14 Thus, it is not surprising to nd a crosssection relationship between political freedom and capitalism, even though no causation is
involved, so there is no inconsistency between the cross-section results and ndings of the analysis
over time.
It is, however, also worth noting the different rates of change in capitalism and political freedom in
the short run. The AdelmanMorris data show that for 17 of the 18 nations in the sample, the
capitalism score increased between 1850 and 1870 and between 1870 and 1890 (Turkey was the one
exception, since its score remained at 0 for the entire 40 year period). The increase in political freedom
was not so uniform. For instance, for the same 18 countries between 1870 and 1890, political freedom
increased in only 11 of the same 18 countries, remained the same in 6 countries, and fell slightly in
one. Between 1850 and 1870 the results were roughly the same. When the composition of nations is
held constant, the average level of political freedom increased slightly. And over the entire nineteenth
century, Table 4 indicates an increase in political freedom in every country except those which started
out with the highest rating.

14
The schooling data come from Barro and Lee (2000). The reported results in the text come from simple linear regressions;
inclusion of per capita GDP did not signicantly improve the results. Inexplicably, the schooling results for 1960 yielded higher
coefcients of determination than the schooling results for 1995 for both political freedom and capitalism. The sample sizes
ranged from 60 (1960 schooling data) to 63 nations (1990 schooling data).

F.L. Pryor / Economic Systems 34 (2010) 91104

103

5. Other issues
There are two methods by which an undemocratic government can maintain its power and deal
with the economy. It can try to manage the economy by itself, a strategy that has the advantage of
maintaining tight political control but risks unimpressive economic growth and rising political
discontent. After World War II, the communist nations of Eastern Europe were able successfully to
maintain both autocratic governments and rising standards of living, at least until the 1980s (when
growth slowed and political discontent increased). Alternatively, an autocratic government can allow
a capitalist economy to develop, which may raise the standard of living but at the possible cost of
nurturing a new business class that may demand greater democracy, or political power for itself. This,
of course, is the reverse of the causation that Friedman posited, but the possibility deserves
consideration.
The fact that in 2000, as well in 1850, 1870, and 1890, there were a number of countries with low
political freedom but a high degree of capitalism means that many non-democratic governments have
selected this option. The key question is whether a government can manage such an arrangement
without undermining its political dominance.
Currently, some undemocratic modern nations appear to have been able to maintain this balance,
at least up to now.15 Although political freedom has appeared to have inexorably increased over the
last two centuries, some doubt remains about whether this trend will continue. Political scientists,
such as Kagan (2008), argue that as long as an autocratic government produces a rising standard of
living, the population can be bought off and there will be no strong push for greater democracy, as we
see in Russia, China, and Singapore today. The Chinese case is particularly interesting and
controversial. Some, such as Tsai (2007), nd little support among Chinese businesspeople for
democratization: by steering carefully among the politicians, they can gain almost everything they
want so they do not need or desire any major changes in the status quo. Those in other segments of
Chinese society appear, at least at present, to have little power to change the situation. Even as China
achieves a more capitalist economy and higher standards of living, those who are too insistent in their
demands for greater democracy may lose their jobs or occupational license or go to prison. Other
specialists writing in Li (2008) reject this view of China and nd some evidence for optimism about
increasing democracy in China.
Unfortunately, this issue about Chinas democratization cannot be resolved here. Nevertheless, if
these modern autocratic governments continue to achieve higher standards of living for their people,
we should not be condent that the democratization trends found in other countries will occur in
these currently non-democratic nations as well.
6. Summary
This essay focuses on Milton Friedmans conjecture that capitalism is a necessary, but not
sufcient, condition for political freedom. I provide evidence for three main propositions:
(1) A cross-section calculation of 90 countries for the decade around the year 2000 shows a strongly
positive relationship between modern capitalism and political freedom. The same results are
found for three cross-sections of countries in the nineteenth century. However, later analysis
shows that no causal connection between the two variables in either direction seems apparent.
Among economically developed nations, the degree of capitalism also appears unrelated to the
length of time that the country has been capitalist. This arises because of little difference between
their capitalism scores.
(2) In the nineteenth century, a number of countries had a relatively high degree of political freedom
but were not capitalist, contrary to Friedmans hypothesis. Moreover, I could nd no evidence that
the change in the political freedom score was related to the change in the capitalism score.
(3) In the nineteenth century, capitalism did not necessarily arise in nations with greater political
freedom than other nations.
15

Besley and Kudamatsu (2008) present a useful analysis of economies with autocratic governments.

104

F.L. Pryor / Economic Systems 34 (2010) 91104

In brief, although political freedom and capitalism are correlated on a cross-section basis,
capitalism is neither a necessary nor sufcient condition for political freedom. And, I might add, I could
nd no conrmation for the reverse proposition either, that political freedom is a necessary or
sufcient condition for capitalism.
Acknowledgements
I would like to thank Thomas Dee, the late Zora Pryor, David Smith, Victoria Wilson-Schwartz, and
members of the Tri-College Brown Bag Lunch Seminar for useful comments on a preliminary version of
this essay.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/
j.ecosys.2009.09.003.
References
Acemoglu, D., Johnson, S., Robinson, J.A., Yared, P., 2008. Income and democracy. American Economic Review 98, 808842.
Adelman, I., Morris, C.T., 1978. Patterns of market expansion in the nineteenth century: a quantitative study. In: Dalton, G. (Ed.),
Research in Economic Anthropology, vol. 1. JAI Press, Greenwich, CT, pp. 231325.
Barro, R., Lee, J.-W., 2000. International Data on Educational Attainment: Updates and Implications. CID Working Paper 042,
Cambridge, MA. Available at: www.cid.harvard.edu/ciddata/ciddata.html.
Beach, W.W., Kane, T., 2008. Methodology: Measuring the 10 Economic Freedoms. Available at: http://www.heritage.org/Index.
Besley, T., Kudamatsu, M., 2008. Making autocracy work. In: Helpman, E. (Ed.), Institutions and Economic Performance. Harvard
University Press, Cambridge, MA, pp. 452511.
Britnell, R.H., Campbell, B.M.S. (Eds.), 1995. A Commercialising Economy: England 1086 to c. 1300. Manchester University
Press, Manchester.
Center for International Development and Conict Management, 2008. Polity IV Project. Available at: www.cidcm.umd.edu/
polity.
Chang, H.-J., 2005. What is wrong with the ofcial history of capitalism? In: Fullbrook, E. (Ed.), A Guide to Whats Wrong with
Economics. Anthem Press, London, pp. 279288.
Freedom House, 2008. Freedom in the World. Available at: www.freedomhouse.org.
Friedman, M., 1962. Capitalism and Freedom. University of Chicago Press, Chicago, IL/London.
Friedman, M., 1991. Economic Freedom, Human Freedom, and Political Freedom. Smith Center for Private Enterprise Studies.
Available at: http://www.cbe.csueastbay.edu/sbesc/frlect.html.
Gwartney, J., Lawson, R., Sobel, R.S., Leeson, P.T., 2007. Economic Freedom of the World: 2007 Annual Report. Available at:
www.freetheworld.com/datasets_efw.html.
Hayak, F.A., 1944. The Road to Serfdom. University of Chicago Press, Chicago, IL.
Heritage Foundation, 2008. 2008 Index of Economic Freedom. Available at: http://www.heritage.org/Index.
Kagan, R., 2008. The Return of History and the End of Dreams. Knopf, New York.
Kaufmann, D., Kraay, A., Mastruzzi, M., 2007. Governance Matters VI: Aggregate and Individual Governance Indicators. World
Bank Policy Research Working Paper 4280, Washington, DC. Available at: www.worldbank.org.
Li, C. (Ed.), 2008. Chinas Changing Political Landscape: Prospects for Democracy. Brookings Institution, Washington, DC.
Maddison, A., 2003. The World Economy: Historical Statistics. OECD, Paris (CD Rom).
Marshall, M.G., Jaggers, K., 2005. Polity IV Project: Political Regime Characteristics and Transitions, 18002004. Available at:
www.cidem.umd.edu/polity.
Morris, C.T., Adelman, A., 1988. Comparative Patterns of Economic Development. Johns Hopkins University Press, Baltimore,
MD, pp. 18501914.
Pryor, F.L., 1972. An international comparison of concentration ratios. Review of Economics and Statistics 54, 130140.
Pryor, F.L., 2005. Economic Systems of Foraging, Agricultural, and Industrial Societies. Cambridge University Press, New York.
Pryor, F.L., 2009. Web site: www.swarthmore.edu/SocSci/Economics/fpryor1.
Rothschild, K., 2003. Reections on an anniversary: Friedmans capitalism and freedom. Journal of Economic Studies 30,
548557.
Sirageldin, I.A.-H., 1969. Non-market Components of National Income. Institute for Social Research, University of Michigan, Ann
Arbor, MI.
Tsai, K.S., 2007. Capitalism Without Democracy: The Private Sector in Contemporary China. Cornell University Press, Ithaca, NY.
World Bank, 2008. World Development Indicators. World Bank, Washington, DC Available at: www.worldbank.org.