Democracy, Technological Change and EconomicGrowth

Carl Henrik KnutsenDepartment of Political Science, University of Oslo

Contact: c.h.knutsen@stv.uio.no

April 13, 2010

Abstract

There is no consensus among academics or policy makers on whether or how democ-racy affects economic growth. This paper provides solid evidence for the hypothe-sis that democracy enhances growth. The paper reviews and evaluates some of themost important theoretical arguments on why democracy and dictatorship may af-fect growth. Particularly, this paper investigates how regime type affects technologicalchange, arguably the most important determinant of long term economic growth. Onehypothesis is that democracies have higher technology-induced economic growth thandictatorships. This hypothesis is deduced from a formal model where dictators valueboth personal consumption and staying in office. In the model, dictators can restrictcivil liberties and diffusion of information, which reduces economic growth, but in-creases dictators’ probability of surviving in office. Thereafter, the paper presents themost extensive statistical study conducted on the effect from democracy on technologi-cal change and on economic growth, drawing on panel data with time series going backto the nineteenth century for some countries. The analysis finds that democracy mostlikely produces higher technology-induced economic growth. There is also relativelyrobust evidence for the hypothesis that democracy enhances economic growth more ingeneral.

1

1 Introduction

Does democracy enhance economic growth?1 The ”Lee-thesis” (Sen 1999, 15), credited to

former Singaporean PM Lee Kuan Yew, postulates that particularly in developing countries,

a strong authoritarian regime is necessary to promote economic development. The East

Asian Tiger-states, Pinochet’s Chile and present-day China are considered decisive empirical

evidence for this assertion. However, several analysts have to the contrary argued that

democracy is beneficial for economic growth, claiming a ”democracy advantage” (Halperin,

Siegle and Weinstein 2005). Figure 1 indicates this may be the case. It shows the smoothed

3-year average GDP per capita growth rate for relatively democratic and dictatorial countries

from 1852 to 2003. Dictatorships have, on average, very seldom outgrown democracies with a

large margin. Rather, democracies have on average mostly had about equal or higher growth

rates than dictatorships, despite the dramatically changing composition of the democracy-

club over the period. However, to investigate whether there is indeed a positive effect from

democracy on growth, we must conduct more stringent analysis.

Although there are adherents to both the Lee-thesis and the democracy advantage view,

and the latter finds strong support in recent statistical analyses (Leblang 1997; Tavares and

Wacziarg 2001; Baum and Lake 2003; Bueno de Mesquita et al. 2003; Papaioannou and

Siourounis 2008; Feng 2005; Doucouliagos and Ulubasoglu 2008), the ”agnostics” have been

prominent. Two seminal contributions are Przeworski and Limongi (1993) and Przeworski

et al. (2000). The first surveyed earlier statistical studies, and found that studies claim-

ing positive, negative or no effect from democracy on economic growth were about equal

in numbers. The second conducted a thorough statistical study of its own, and found no

robust effect from democracy on GDP growth. However, the study does find indications of

1Democracy is here defined as a political system with high degrees of popular control over public decisionmaking and political equality(see e.g. Beetham 1999). Free and fair elections are crucial for securing popularcontrol over politics, but also political and civil rights are essential for democracy according to this definition(Knutsen 2011).

2

Figure 1: The figure shows smoothed three-year average GDP per capita growth for relativelydemocratic (Polity-index≥ 6) and dictatorial countries from 1852 to 2003.

3

a positive effect from democracy on GDP per capita growth. Also other studies confirm the

agnostic position (e.g. Burkhart and Lewis-Beck 1994; Helliwell 1994). These results are ac-

cepted by many prominent political scientists. Diamond (2008, 96), for example, asserts the

”evidence is murky” for the hypothesis that democracy spurs economic development, while

Tsebelis (2002, 70) reports it as a surprising fact that there is no evidence of democracies

producing superior economic outcomes. However, two points should be noted: First, studies

reporting insignificant effects from democracy on growth have often studied only the direct

effect, as they have controlled for variables constituting the main channels through which

democracy affects growth. Tavares and Wacziarg (2001), Baum and Lake (2003), Feng (2005)

and Doucouliagos and Ulubasoglu (2008) find that democracy has a positive total effect on

growth, for example because it enhances factors like human capital accumulation. Second,

the time dimension and other sample characteristics influence results (see Przeworski and

Limongi 1993; Doucouliagos and Ulubasoglu 2008). Doucouliagos and Ulubasoglu (2008)

find that studies with longer time dimensions and more data points are more reliable. In

addition to short time series affecting results, many dictatorships with poor economic track-

records, like North Korea, Myanmar, Cuba and Eritrea, are often either unwilling or unable

to provide infrastructure for data collection (see Knutsen 2011). However, both fast-growing

dictatorships and almost all democracies provide data (Halperin, Siegle and Weinstein 2005,

32-33).

This study incorporates the largest number of country-year observations in the litera-

ture, including more than 150 countries with time series going from 1820 to 2003 for some

countries. Moreover, this study investigates the total, rather than the direct effect from

democracy on growth, using pooled cross-section - time series and panel data techniques.

The effect from democracy on growth is relatively robust, but not completely. All in all,

however, the results presented here constitute a compelling case for a positive effect from

democracy on growth. Moreover, this paper elaborates on one important channel through

4

which democracy likely enhances economic growth, namely technological change. The previ-

ous literature on democracy’s economic effects has mostly focused on how democracy affects

growth through physical- (see e.g. Przeworski and Limongi 1993; Tavares and Wacziarg 2001)

and human capital accumulation (see e.g. Baum and Lake 2003; Stasavage 2005).2 The eco-

nomic growth literature, however, considers technological change the central determinant of

long run growth (see e.g. Solow 1957; Nelson and Winter 1982; Romer 1990; Aghion and

Howitt 1992; Grossman and Helpman 1991; Klenow and Rodriguez-Clare 1997; Helpman

2004; Nelson 2005) (but see Mankiw, Romer and Weil 1992). A model showing how self-

interested dictators restrict civil liberties for political survival purposes is presented. In an

imperfect world where dictators can not fine-tune policies, this also inhibits dissemination

of economically relevant ideas and technologies. Limitations on freedom of speech, media

and travel inhibit economic actors’ from freely learning and adopting foreign and nation-

ally developed ideas. Thus, dictatorship slows technological change. This hypothesis finds

relatively robust empirical support when using Total Factor Productivity (TFP) growth as

proxy for technological change.

2 Literature review

In their seminal study, Przeworski and Limongi (1993) evaluated four arguments on the rela-

tionship between regime types and growth. The arguments highlight how regime type affect

property rights, investment, state autonomy and checks on predatory rulers. I reevaluate

these arguments, and include some new theoretical insight and empirical results. There-

after, I present other important arguments, particularly on democracy and human capital

accumulation.

2There are some exceptions. Przeworski et al. (2000) conducted growth accounting on data from 1950 to1990, and their results indicated that democracies may do better on technological change, but only amongrich countries. Pinto and Timmons (2005) also investigated the relationship, but relied on problematicproxies of technological change, like foreign direct investment and trade.

5

2.1 Property rights

Przeworski and Limongi (1993) assess the nineteenth century-debate on democracy’s eco-

nomic consequences; the right to vote and freedom of organization were widely perceived

to have adverse effects on private property rights protection, and thus economic growth.3

Political economic models (see e.g. Meltzer and Richards 1981; Boix 2003; Acemoglu and

Robinson 2006b) indicate more progressive property redistribution in democracy, as the me-

dian voter is assumed to be poorer than those who set economic policy under dictatorship.

Poor voters, powerful because of their numbers under democracy, are expected to use their

political power to redistribute property from the rich. However, the threat to property rights

may not necessarily come from the poor; redistribution is not necessarily progressive. Any

form of government implies concentration of coercive power, which opens up for state-led

confiscation from rich and poor. However, in democracies the government’s supporters are

many, and thus internalize much of the negative effects from property rights violation on

incentives for productive activity (Olson 1993). Second, there is more power dispersion in

democracies, also between state institutions. Because of poor political accountability and

concentration of power, dictatorial elites can confiscate property more easily. Third, re-

distribution of property as ”private goods” to political supporters is relatively cheaper in

dictatorships, where the rulers’ ”winning coalitions” are smaller (Bueno de Mesquita et al.

2003). Under democracy, where winning coalitions are larger, rulers motivated by political

survival will have greater incentives to provide the ”public good” of universal property rights

protection. Several empirical studies find a positive effect from democracy on property rights

protection (Leblang 1996; Adsera, Boix and Payne 2003; Clague et al. 2003; Knutsen 2010a).

3See e.g. North (1990); Knack and Keefer (1995); Acemoglu, Johnson and Robinson (2001) for theimportance of property rights for economic growth.

6

2.2 Physical capital

There may be a dictatorial advantage in physical capital accumulation (see e.g. Przeworski

and Limongi 1993; Przeworski et al. 2000; Tavares and Wacziarg 2001; Knutsen 2010b), as

dictatorship allegedly increases savings rates. First, dictatorships often suppress freedom of

association, thus crippling independent organization of unions. In the absence of unions,

wages are likely lower, and capital-owners earn more. Thus, if savings rates increase with

income (the Kaldor-hypothesis), aggregate savings rates will be higher in dictatorships. Also,

dictatorships lack free and fair elections, which reduces the pressure on leaders to channel

resources to public consumption. Instead, dictators can push different savings-enhancing

policies, like banning luxury consumption or restricting consumer loans (see e.g. Wade 1990;

Chang 2006), independent of the desires of ”short-sighted electorates”. Likewise, dictatorial

governments can more easily reduce social security spending; the rational response of citizens

is to save privately to self-insure for the future. However, empirically there are relatively few

dictatorships with extremely high savings- and investment rates; some exceptions are the

Soviet Union, the East Asian Tigers and present-day China. The correlation between the

Polity-index and the investment/GDP ratio (from World Development Indicators), based on

3517 country-years from 1960 to 2004, was actually slightly positive (.02). Self-interested

dictators may not want policies that lead to investment-induced growth, as becomes clear

from the argument below on predatory dictators. Investment, and particularly foreign direct

investment (FDI), is sensitive to property rights protection, where democracies were argued

to have an advantage (see Li and Resnick 2003). Thus, some studies find indications that

democracy increases foreign direct investment (FDI) (e.g. Busse and Hefeker 2007). This

partly mitigates dictatorship’s capital advantage from higher domestic savings rates.

7

2.3 State autonomy

Scholars studying East Asia have linked some Asian dictatorships’ economic performance

to the dictatorial state’s autonomy; ”the key to the superior economic performance of the

Asian ”tigers” is ”state autonomy,” defined as a combination of the ”capacity” of the state

to pursue developmentalist policies with its ”insulation” from particularistic pressures, par-

ticularly those originating from large firms or unions. This argument takes two steps: ”state

autonomy” favors growth, and ”state autonomy” is possible only under authoritarianism”

(Przeworski and Limongi 1993, 56). Olson (1982) argues that democracies are prone to cap-

ture by special interest groups. This may lead to policies that sacrifice economic growth for

the protection of specific business sectors or pivotal voters whose interest is not aligned with

growth. Microeconomic reforms or trade reforms may improve efficiency, but certain privi-

leged groups may lose out. Under democracy, these potential losers may be ”veto players”

(Tsebelis 2002), and block reform. Allegedly, politicians and bureaucrats in dictatorships

are often insulated from such pressures, and thus better able to conduct ”proper” policies

(e.g. Wade 1990; Leftwich 2000). Reform may also be enacted quicker under dictatorship,

since procedural steps and time-consuming negotiation can be skipped. However, it is ques-

tionable whether dictators are as autonomous as described above. Despite lack of free and

fair elections, no dictator survives without political supporters, be it the party, landowners

or the military. Bueno de Mesquita et al. (2003, 7) recognize that every leader ”answers to

some group that retains her in power: her winning coalition”. The winning coalition is again

drawn from a ”selectorate”, the group of actors that can influence the selection of leaders.

The difference between democracies and dictatorships is therefore not regimes’ degree of

autonomy, but the sizes and nature of the underlying winning coalitions and selectorates.

This insight may lead us to rethink the supposedly positive economic effects of dictatorial

insulation from the general populace.

8

2.4 Predatory rulers

According to Evans (1995, 45), if autonomy is defined as not having goals shaped by social

forces, Mobutu’s regime in Zaire was very autonomous. Mobutu’s is an infamous example of

a ”predatory regime”, where the dictator and his inner clique used their powers for enriching

themselves and securing continuation in office. Dictators often promote policies to their

own benefit, even if the population suffers economically (see e.g. Robinson 1998; Bueno de

Mesquita et al. 2003; Acemoglu and Robinson 2006a; North, Wallis and Weingast 2009).

Historically, examples vary from Louis XIV’s Versailles-project to Pol Pot’s decision to kill

Cambodians with education or glasses. Clear-cut examples of predatory rulers are those that

confiscate socially productive resources for personal consumption. More generally, rulers

often use strategies for achieving personal goals, but which may reduce economic growth.

Many dictators want to minimize the probability of being thrown out of office. If an office-

motivated dictator for example believes that industrialization leads to a strong middle class

or organized working class calling for democracy, the dictator will be better off not promoting

industrialization. In democracies, leaders who try to engage in predatory activities are more

likely to be detected because of free media, more likely to be checked by the legislature and

courts, and are likely thrown out of office in the next election. These institutional features

thus provide checks on predatory behavior. However, not all dictatorships are predatory. One

reason is that dictatorships vary in terms of institutionalized checks and balances (Przeworski

et al. 2000). Moreover, in some contexts, self-interested, rational dictators may refrain from

predatory activities, because future income is tied with the economy’s growth (Olson 1993)

or because they may lead to revolution (see e.g. Overland, Simmons and Spagat 2000).

Bueno de Mesquita et al.’s (2003) analysis indicates that if dictators have relatively large

winning coalitions, they will have incentives to provide public goods instead of engaging

in predatory behavior. (Besley and Kudamatsu 2007) argue that winning coalitions who

can retain their positions as political players if a particular dictator falls from power can

9

discipline the dictator.

2.5 Human capital and other channels

More recently, several studies have argued that democracy enhances growth through in-

creasing human capital (Tavares and Wacziarg 2001; Baum and Lake 2003; Doucouliagos

and Ulubasoglu 2008). (Lindert 2005) finds a strong positive effect from franchise expansion

on educational expansion in Western Europe. Franchise expansion increased politicians in-

centives’ to provide better education quality and broader coverage (see also Acemoglu and

Robinson 2006b). L., Mariscal and Sokoloff (1998) find the same effect in Latin America.

Stasavage (2005) finds that democracy has a positive effect on primary education spending

in Africa. Broadening education coverage is often an inefficient way for dictators to buy

political support, but a relatively cheap way of buying support for democratic leaders with

large winning coalitions (Bueno de Mesquita et al. 2003). Democracy may affect growth

also through other channels: Feng (2005) argued that democracy affects growth positively

through enhancing political stability. Also, democracy likely reduces corruption, at least in

consolidated democracies (e.g. Rock 2009), and corruption affects growth negatively (Mauro

1995). Moreover, Rodrik (1999) argues that democracies may be better able to implement

politically difficult, but economically efficient, reforms; the reason is that democracies posit

institutions that enable compensation of reform-losers. As Acemoglu and Robinson (2006b)

and Wintrobe (1990) point out, dictatorships also spend more resources on repressive appa-

ratuses, which takes away resources from other productive tasks.

In conclusion, except for dictatorship enhancing physical capital accumulation and per-

haps contributing to efficient but unpopular reforms in particular contexts, most arguments

above indicate an economic growth advantage for democracy. There is however another,

surprisingly overlooked, channel through which democracy may enhance economic growth,

technological change.

10

3 Technological change

3.1 Technological change and economic growth

Historically, macroeconomists emphasized accumulation of inputs as determinants of in-

come (Helpman 2004, :9). However Abramowitz (1956), Solow (1957) and Denison (1962)

found that technological change contributed more than input accumulation to growth in

the US, and Denison (1968) found the same pattern for European countries. Later, (e.g.

Easterly 2001) has argued that technological change is key to growth also for poorer coun-

tries. Some empirical estimates indicate that differences in technological efficiency explain

the bulk (about 9/10) of cross-national variation in income globally (Klenow and Rodriguez-

Clare 1997). Romer (1993) discussed the importance of open idea flows for economic growth.

Ideas are non-rivalrous entities (Romer 1993); an idea can be used by several actors without

diminishing its value for others. Technological change thus not only contributes to growth

in rich countries at the ”technological frontier”, but also in developing countries (Helpman

2004), as poorer countries can adopt technological (and organizational) improvements devel-

oped elsewhere. Understanding why some countries are better at adopting production- and

organization techniques, and diffusing them throughout their economies, is therefore crucial

for understanding income level and growth differences. 4

In the model below, self-interested dictators restrict information flows by curbing civil

liberties. This, in turn, reduces technological change. The model is part of a larger class

of models, where self-interested dictators may have incentives to take actions with negative

consequences for their national economies. For example, in Olson (1993), dictators, especially

those with short time horizons, expropriate property to maximize personal consumption,

thereby reducing incentives for citizens to work or invest. In Robinson (2001) and Acemoglu

4The separation of growth due to human and physical capital accumulation and technology is not straight-forward. For example, new technologies often come with new investments in machinery (Nelson 2005), anda high human capital level may be conducive to the spread of more efficient technologies (Kremer 1993).

11

and Robinson (2006a), dictators also maximize discounted utility from consumption. In these

models, public investment and economic development more generally strengthen opposition

groups and reduce leaders’ survival probability. Leaders may thus shrink their economies,

among others through manipulating public investment levels, to maximize expected utility

from (present and) future private consumption. Bueno de Mesquita et al. (2003) assume

political leaders are interested in surviving in office, and shows how dictators under-invest in

growth-conducive public goods. For these leaders, it is rational to expropriate or tax heavily

and redistribute resources as private goods to their relatively small winning coalitions. In

Wintrobe (1990), power-motivated dictators overinvest in repressive capacity, which distorts

public resources away from more productive projects. Before presenting the model, I expand

on the role of civil liberties and open information flows for technological change.

3.2 Civil liberties, information flows and technological change

Civil liberties are better protected in democracies than in dictatorships (see e.g. Beetham

1999). Dictatorial elites may because of limited knowledge or self-interest suppress ideas

that are essentially correct(see e.g. Mill 1974), and these ideas may be of both economic

and political relevance. Open debate and free idea flows are crucial for efficient decision-

making by firms, bureaucrats and politicians, as ”the knowledge of the circumstances of

which we must make use never exists in concentrated or integrated form, but solely as the

dispersed bits of incomplete and frequently contradictory knowledge which all the separate

individuals possess” (von Hayek 1945, :519). A broad diversity of ideas improve economic

efficiency, especially when economic actors easily learn of the new ideas and select the most

efficient. A dynamic economy ”is the outcome of a constant interaction between variety

and selection” (Verspagen 2005, :496). Civil liberties, such as freedom of speech, press and

travel allow for introduction of new ideas and idea diffusion. Civil liberties also allow for

comparison of different ideas, thus allowing for the more efficient to win out. Civil liberties

12

therefore enhance both variety and selection, as the introduction of new ideas, but also

learning processes, rely on the possibility of collecting and processing information in a fairly

unrestricted manner.

Efficiency change comes not only from product innovations, but also from introduction

of new policies and changes in economic institutions and organizations(North 1990). North

(2005) argues the inherent uncertainty about policy- and organizational effects necessitates a

process of trial and error, with proper feedback from society. Open systems, associated with

democratic government and civil liberties, are crucial for such processes. In other words,

”open access orders more readily generate a range of solution to problems; they more readily

experiment with solutions to problems; and they more readily discard ideas and leaders who

fail to solve them” (North, Wallis and Weingast 2009, :134). The opportunity for actors

outside government to freely opine on political reforms therefore improves organizational-

and policy efficiency. 5 Restrictions on freedom of speech and media hurt efficiency, as

important problems are not reported and alternative viewpoints on economic policies are

either not forthcoming to the political rulers or properly debated. Civil liberties, but also

open political competition among self-interested elites (see North, Wallis andWeingast 2009),

ensure that various ideas are put forth, and properly debated. This enhances efficiency in

democracies. In contrast, dictators often seek to limit idea variety in order to retain power.

Democracies should thus have higher long-run growth rates due to more rapid technological

change.

Good examples of the mechanisms above relate to communication technologies: ”The

Internet presents a dilemma to leaders of authoritarian states and illiberal democracies. It

promises enticing commercial advantages, such as transaction cost reductions, e-commerce

possibilities, and foreign trade facilitation. Yet, by giving citizens access to outside infor-

5The ”dictator’s dilemma” is also relevant (Mueller 2003, 416-7). Because of fear of falling out with thedictator, individuals and organizations may not be forthcoming with their most accurate information, whichreduces decision making quality.

13

mation and platforms for discussion and organization, the Internet can also help politically

empower populations and potentially threaten regimes” (Hachigian 2002, :41). Cell-phone

technology also present both political problems and potential economic gains to dictators.

Bans on cell-phones have been imposed in Cuba and Turkmenistan, for example. Histori-

cally, one dictatorial regime generating technological and economic stagnation was imperial

Quing-China. China experienced a dramatic relative economic decline compared to West-

ern Europe, especially from the 19th century (see e.g. Landes 2003). The Chinese empire

was characterized by the ruling dynasty’s concentration of power and its closed nature, in

terms of foreign relations. Political rulers neglected and even outlawed new and more ef-

fective organization techniques and production technologies. Especially foreign ideas were

espoused. A letter from Emperor Ch’ien Lung to George III of England in 1793, illustrates

that dictatorship can be obstructive to adaptation of foreign ideas: When denying the British

trade requests, the Chinese Emperor wrote that the ”Celestial Empire possesses all things

in abundance. We have no need for barbarian products” (Murphey 2000, :245).

3.3 A political economic model of information flows and techno-

logical change

In the model below, self-interested dictators may have incentives to conduct policies with bad

macroeconomic effects. The model is fairly simple, as the focus here is relatively specific; the

model shows how dictators have incentives for restricting information flows, thus crippling

diffusion of technology. The model thus focuses on political institutional characteristics that

”frame the struggle between the proponents of change and their opponents, and thereby

affect the ability of countries to innovate and to implement new technologies” (Helpman

2004, :112).

14

3.3.1 The economy

We use an adjusted neo-classical production function, as in (Mankiw, Romer and Weil 1992):

Y = F (TL,K,H), where Y is output, T technology level, L labor input, K physical capital

input and H human capital input. F is increasing, but concave, in all inputs. ∂Y∂T

= L · ∂F∂TL

>

0: Output increases in technological efficiency. For simplicity, we use a Cobb-Douglass

specification:

Y = F (K,L,H, T ) = KαHβ(TL)1−α−β. (1)

Technology is endogenous. However, this model does not analyze firms’ incentives to

generate new technology as in ”new growth theory” (Romer 1990; Grossman and Helpman

1991; Aghion and Howitt 1992). Generation of cutting-edge technology in increasing-returns-

to-scale sectors is mostly relevant for large and rich developed countries. However, for most

countries the global technological frontier is largely exogenous, and the diffusion (and local

adaptation) of international technology is key for technological efficiency. Thus, one can

focus on technology diffusion when modeling cross-country differences in technology-induced

economic growth. In the model, national technological change is a function of how many new

techniques national economic actors adopt, denoted At. More specifically, the rate of change

in technology is TT= ω(At). The number of new techniques developed each year globally is

A∗t , and treated as exogenous. In accordance with the discussion above, national information

flows, i, determine the degree to which a country utilize new, globally developed ideas to

produce technological change. i comes in two pure types, politically- and economically

relevant information, ie and ip. However, there is also non-pure information, iep, of both

economic and political relevance. Only ie and iep affect technological change. We thus have

that, At is a function of A∗t , ie and iep. We normalize so that ie + iep varies between 0 and

1, with 0 indicating a country that restricts all economic information flows and 1 indicating

a country that allows for the free flow of economic information. We assume, in the simplest

15

of models, that At = (ie + iep)A∗t . This means that

T

T= ω((ie + iep)A

∗t ). (2)

It can be shown, through taking logarithms and differentiating the production function, that

Y

Y= (1− α− β)

T

T+ α

K

K+ β

H

H+ (1− α− β)

L

L(3)

This again implies that

Y

Y= (1− α− β)ω((ie + iep)A

∗t ) + α

K

K+ β

H

H+ (1− α− β)

L

L(4)

Equation 4 shows that GDP growth rates depend on growth rates of physical capital, human

capital and labor, changes in the global technological frontier and the information flows in

national economies. If countries are in their steady states (see e.g. Barro and Sala-i Martin

2004), income in countries with free information flows will grow with the rate of change in

the global technology frontier. Elsewhere, steady state growth rates will be weighted down

with the degree of information flow-restrictions. A country where almost no information is

allowed, for example North Korea (see e.g. Kihl and Kim 2006), will have very low long-run

growth rates.

3.3.2 Political decision making

Let us endogenize restriction of information flows. First, we simply assume that in democ-

racies, all types of information are allowed.6 Let us therefore consider a dictator, D, in a

two-period model, who maximizes a utility function, U = U(c, q), dependent on personal

consumption, c, and political survival probability in the second period, q. U(c, q) is in-

6The assumption can be weakened to an assumption that democratic leaders restrict civil liberties lessthan dictators.

16

creasing and concave in both arguments. D receives a fixed share, λ, of Y , and therefore,

ceteris paribus, wants to increase the economy’s size to increase personal consumption. D’s

consumption is given by

ct ≤ λYt = λKαt H

βt (TtLt)

1−α−β (5)

Since there is no saving in the model and U ′(c) > 0, 5 holds with equality. We manipulate

the utility function, to analyze how D’s utility depend on consumption growth rate. We

assume an exogenously given Y0, and thus c0, in period 0. Change in consumption, ∆c, is

therefore given by

∆ct = ct − c0 = λKαt H

βt (TtLt)

1−α−β − λKα0 H

β0 (T0L0)

1−α−β (6)

For simplicity, and without loss of generality, we assume that Kt = K0, Ht = H0, Lt = L0, so

that ∆c is only a function of changes in T . Further, if we hold λ constant and use Equation

4, we find that D’s consumption growth rate cc= ∆c

c0, denoted gc, is given by

gc = (1− α− β)ω((ie + iep)A∗t ) (7)

Since c0 is exogenous, and U ′(c) > 0, U ′(∆c) > 0, and therefore U ′(gc) > 0.

Information flows are affected by policies such as restrictions on freedom of speech, media

and travel and investment in public goods related to communication technology. These are

the actual policies set by a dictator, but we model their consequence, information flows, as

choice variables to simplify. D sets policy (ie; ip; iep) in the first period. D has a probability

(1− q) of losing power in the second period. Before the revelation of whether D loses power

or not, he receives his income which is used for consumption. We assume that D consumes

whether he loses power or not. q is endogenous to information flows. The probability of

dictatorial survival decreases in ip and iep, but is unaffected by ie. That is∂q∂ip

< 0, ∂q∂iep

< 0

17

and ∂q∂ie

= 0. We model the relationship with the simple, linear function:

q = (1− (γip + ηiep)) (8)

Here, γ > 0, η > 0 and 0 ≤ γip + ηiep ≤ 1;the probability of survival varies between 1 when

no political and mixed political-economic information is allowed, and 0, which results from a

high level of political or mixed political-economic information flow. Generally, it is difficult

for dictatorial governments to screen each act of communication, travel and meeting, and

governments therefore need to establish general rules. Therefore, information activities are

banned under uncertainty of their contents. Civil liberties restrictions may not only reduce

political communication, but also economically relevant communication. Thus, disallowing

general free and open exchange of information and debate will have effects not only in terms

of stifling political opposition, but also economic dynamism. This is captured by iep > 0.One

way to model the relationships between q and ip and iep would be to assume an opposition

consisting of several individuals, all desiring to overthrow D. The probability of the op-

position overthrowing D, (1 − q), depends on coordination. As collective action problems

are solved and opposition-members coordinate, (1 − q) increases. The ability of the oppo-

sition to coordinate depends on their ability to use communication tools, assemble without

harassment or detention, gain access to media and travel freely. Therefore, restrictions on

civil liberties that reduce, ip and iep, reduce the opposition’s ability to coordinate and thus

(1− q).

Let us return to D’s transformed utility function, U(gc, q). If we insert for Equations 7

and 8, we get:

U(gc, q) = U((1− α− β)ω((ie + iep)A∗t ), (1− (γip + ηiep))) (9)

From Equation 9, one may see that D minimizes ip and maximizes ie. D cracks down on

18

all information flows that are politically dangerous but irrelevant for economic efficiency,

and opens up for information that improves economic efficiency but is irrelevant for political

survival. We can show this more stringently:

∂U

∂ip= −γ ∂U

∂q(10)

∂U

∂ie=

∂U

∂gc· (1− α− β)A∗

tω′(At) (11)

Equations 10 and 11 show it is always rational for the dictator to increase ie, as∂U∂ie

> 0 and

reduce ip as ∂U∂ip

< 0. Thereby ie and ip will be set at their maximum and minimum levels

respectively. The interesting trade-off in the model relates to iep. D on the one hand wants

to allow iep because it increases efficiency and thus private consumption growth. But, on the

other, he wants to restrict iep because it puts his political survival at risk. The first-order

condition is given by:

∂U

∂iep=

∂U

∂gc· (1− α− β)A∗

tω′(At)− η

∂U

∂q(12)

Since in optimum ∂U∂iep

= 0, Equation 12 implies that the dictator will set iep, so that:

∂U

∂gc· (1− α− β)A∗

tω′(At) = η

∂U

∂q(13)

7 Equation 13 shows that in optimum, the dictator will balance increase in marginal util-

ity from consumption against the expected marginal utility-decrease from reduced survival

probability, when setting iep. Thus, some iep is restricted in dictatorships, whereas all iep

is allowed in democracies. The model thus indicates that democracies will experience more

rapid technological change than dictatorships.

7This can be rewritten in marginal rate of substitution form to∂U∂gc∂U∂q

= η(1−α−β)A∗

tω′(At)

.

19

Several other propositions follow from the model. For example, dictators ruling over

economies with high capital shares or in periods with slow global technological change, and

dictators with a weak hold on power or who expect reduced consumption after losing power

will restrict civil liberties more harshly. Thus, they slow down technological change relative

to democracies more (see Knutsen, 2011).In Knutsen (2009), I showed how strong state

institutions and an independent, effective bureaucracy are particularly important for growth

in dictatorship, as they curb leaders’ opportunities to promote bad economic policies. This

model presents an additional mechanism explaining the strong interaction between state

capacity and regime type on growth: Even with limitations on freedom of speech and press,

high-capacity dictatorships may perform adequately on idea diffusion, because of better

abilities for absorbing and interpreting weak information signals. Such dictatorships may

also be better able to design policies that enable separation of politically and economically

relevant information. For example, a high-quality bureaucracy may be able to fine-tune

internet policies so that only politically problematic pages are blocked. In terms of the

model, iep decreases as bureaucratic quality, b, increases: iep = iep(b), where i′ep(b) < 0. As

ie + iep = 1, i′e(b) > 0. In democracies, where all information, also all iep, is allowed, b will

not matter for technological change. In a hypothetical case where the bureaucracy is able to

screen perfectly so that iep is zero, democracies and dictatorships have equal technological

change rates. This hypothesis is tested, but find only some empirical support in Knutsen

(2011).

4 Data

Chang (2006, 145) notes that 1960 is ”year zero” in statistical studies on economic devel-

opment. Lacking data has forced quantitative social scientists to leave out the main part of

modern history from the ”dual revolutions”; the industrial revolution in Great Britain and

20

the political revolutions in eighteenth century United States and France. This study is the,

so far, most extensive study on democracy and growth, incorporating data from 1820 (for

some countries) to 2003. There is data for more than 150 countries. Several countries, often

left out of studies because of lacking GDP-data, like North Korea, are included here. This

is important not only because of pure sample-size considerations, but also because results

are biased when the economically worst performing dictatorships are left out. Despite a

preference for the broader Freedom House Index (see Knutsen 2011), I use the Polity-Index

(PI) from the Polity IV data-set, as democracy measure because of its longer time series.

The PI goes from -10 (most dictatorial) to 10 (most democratic). The index’ dimensions are

competitiveness and openness of executive recruitment, constraints on the chief executive,

and competitiveness and regulation of political participation. I leave out country-years in

interregnum-periods (Marshall and Keith 2002, 17), mainly periods of internal anarchy or

civil war. I also leave out country-years coded as ”foreign interruption”, mainly registering

foreign occupation. Unsystematic measurement errors in independent variables generate at-

tenuation biases (Greene 2003, 83-90). If there is unsystematic measurement error in the PI,

the PI-coefficient will be biased towards zero.

Data on GDP and population, collected from Maddison (2006), go back to 1820 for some

countries. Estimating GDP for years when the national accounting system was not invented

is difficult, and Maddison’s estimates contain large errors. Maddison (2007, 294) claims that

”[t]here is still a need to fill gaps and crosscheck existing estimates, but the broad contours

of world development in this period are not under serious challenge” Different sources and

procedures have been used to estimate population and GDP-data (see Maddison 2006, 169-

228). Maddison utilizes PPP-adjusted GDP-data (US 1990$), which take into account local

price levels. I include around 800 country-years where data on GDP per capita growth or level

or population level is constructed by interpolation, assuming constant growth rate. Several

of the time series are interrupted in the Maddison data-set. I only constructed interpolated

21

GDP data where at least 90% of the years missing between two observation-years have the

same PI-score.8 Thus, the dependent variable, GDP per capita growth, is measured with

error. If this error is unsystematic, it will be more difficult to find significant regression

coefficients, for example for the PI.

Measuring technological change is difficult (see e.g. Nelson 2005). One common measure

is TFP-growth. TFP is calculated as a residual, where economic growth stemming from

changes in physical capital, human capital and labor are subtracted from total growth (see

e.g. Barro and Sala-i Martin 2004; Baier, Dwyer Jr. and Tamura 2002). I utilize the extensive

TFP-data from Baier, Dwyer and Tamura (2006), covering 145 countries, with 24 countries

having time series longer than 100 years. The TFP-data are estimated with uneven intervals,

approximately averaging a data point every tenth year for most countries.9 I interpolate these

series, assuming constant TFP-growth rates within periods. I use both the interpolated

annual data, and the periodic observations given by Baier, Dwyer and Tamura (2006) in

different analyses below. There are several potential biases in TFP (see e.g. Rodrik 1997;

Verspagen 2005). Among others, it may be underestimated since investment likely increases

when technology level increases. If so, technological change is a cause of capital accumulation,

but growth will be assigned to capital accumulation in the growth accounting. Nevertheless,

there is no obvious reason why such biases should critically influence the relationship between

democracy and TFP. Measurement error may be a particular problem for the older data

(see e.g. Maddison 2006; Baier, Dwyer and Tamura 2006). If TFP measurement errors are

unsystematic, they do not bias regression coefficients but only increase their standard errors,

thereby making it harder to find significant results.

Technological change and economic growth are not only a function of regime type. I

8I thus have artificially low variation in the estimated GDP per capita growth rates, but the averagegrowth rate is correct by construction.

9These authors use data from multiple sources, and calculate TFP using income per worker rather thanper person. They assume Hicks-neutral technology and a capital share of 1/3. For a closer description ofthese data and the underlying assumptions, see Baier, Dwyer Jr. and Tamura (2002).

22

thus control for several variables. Because of probable convergence effects, log TFP-level is

controlled for in TFP-growth regressions, and log GDP per capita-level in GDP per capita-

growth regressions. Log population is controlled for in both regressions, and so is log regime

duration, a proxy for political stability based on data from Polity IV. Moreover, I control for

ethnic fractionalization, using data from Alesina et al. (2003). To control for cultural and

political-historical factors, plurality religion and historical colonizer dummies are entered.10I

also add region- and decade dummies to control for geographic- and time-specific effects.

Papaioannou and Siourounis (2008) find that democracy’s effect on growth peaks and sta-

bilizes after 3 years. However, there are reasons to believe that the lag in effect from regime

type on TFP-growth in particular is longer Knutsen (2011). We thus run regressions that

use a 5-year lag on all independent variables. Moreover, the longer the PI-lag, the smaller

are potential endogeneity problems related to growth rates influencing regime type (Helli-

well 1994) (also remember that we control for income level). However, robustness checks are

conducted with 2- and 3-year lags.

5 Empirical analysis

5.1 Technological change

I run OLS with Panel Corrected Standard Errors (PCSE) (see Beck and Katz 1995), which

takes into account heterogeneous standard errors and contemporaneous correlation between

panels, and AR(1) autocorrelation within panels. OLS with PCSE allows us to draw on

information from both cross-country and within-country comparisons. However, there may

be country-specific characteristics biasing these results. I therefore run RE and FE to check

the results’ robustness. The models in Table 1 utilize the interpolated TFP-data, with

country-year as unit of analysis. There is generally very good support for the hypothesis that

10See Knutsen (2007, 67-68) for the coding of these dummies.

23

democracies experience more rapid technological change; dictatorship reduces TFP-growth

relative to democracy. In all models, the regime coefficient has the expected sign, and the

effects are all significant at the 1%-level. The t-values for the RE- and FE models with

the 5-year lag are above 5, indicating that democracy clearly enhances technology-induced

growth, even when taking into account country-specific factors. The size of the estimated

effects are quite substantial; they indicate an effect of about 0.6 to 0.7 percent extra TFP-

growth per year, when going from most dictatorial to most democratic. When taking a long

view, say 100 years, a country with a 0.7 percent higher TFP-growth than another otherwise

equal country, would be twice as rich as the other at the end of the period, if starting out

equally rich. If democracies in addition grow more because they enhance human capital

accumulation (see e.g. Tavares and Wacziarg 2001; Baum and Lake 2003), democracies have

a quite substantial growth advantage on dictatorships because of knowledge-related factors.

Dictatorships may hoard physical capital investment better (Tavares and Wacziarg 2001;

Przeworski and Limongi 1993), but in the long run knowledge matters, and democracies

prosper.

The results above were robust to substituting the 5-year lag with a 3-year lag.11 However,

regressions using 2-year lags yielded insignificant results. I also ran models using 10-year lags,

and models using Freedom House’s Civil Liberties index (data back to 1972) as independent

variable. The results were relatively robust to these changes. The results were also relatively

robust to controlling for other factors, such as absolute latitude and the Frankel-Romer

trade index. Moreover, the model above indicated that dictators’ survival probability, and

thus regime duration, is endogenous to technologically relevant policies. We thus left out

log regime duration, but the results were very similar, although some models showed even

larger PI-coefficients and t-values. However, the interpolation conducted above may be

problematic, as it expands the number of data points and introduces additional measurement

11All robustness check-results are available on request.

24

Table 1: OLS with PCSE, RE and FE models with 5-year lags and TFP-growth as dependentvariable. Time dummies and constant omitted from table. *** p < 0.01, ** p < 0.05, *p < 0.10

Variable—Model OLS PCSE RE FEb (t) b (t) b (t)

Polity 0.029*** (3.32) 0.037*** (5.59) 0.034*** (5.03)Ln TFP -2.031*** (-3.64) -2.679*** (-17.62) -3.241*** (-20.32)Ln Population -0.106 (-1.33) -1.222*** (-11.81) -2.634*** (-17.85)Ln Regime dur. 0.013 (0.50) -0.113*** (-3.93) -0.144*** (-5.02)Ethnic fraction. -1.002** (-1.97) -1.214 (-1.33) .British 0.103 (0.32) -0.322 (-0.59) .French 0.164 (0.63) -0.394 (-0.61) .Portuguese -0.836 (-0.94) 0.406 (0.35) .Spanish 0.321 (0.43) 0.029 (0.03) .Belgian -2.905** (-2.45) -3.382** (-2.17) .Sunni -0.494 (-0.51) 0.271 (0.14) .Shia -1.581 (-1.37) 0.525 (0.23) .Catholic 1.197 (1.00) 2.078 (1.00) .Protestant+ 0.605 (0.57) 0.842 (0.41) .Orthodox -1.296 (-1.19) -0.759 (-0.35) .Hindu -0.121 (-0.11) 0.422 (0.18) .Buddhist+ 0.619 (0.57) -0.528 (-0.24) .Indigenous -0.903 (-0.84) -0.952 (-0.44) .E.Eur.+Soviet rep. -1.830*** (-3.71) -4.339*** (-5.80) .Africa SS -1.306** (-1.97) -3.003*** (-3.58) .Asia-Pacific -1.313* (-1.68) 0.040 (0.04) .MENA 1.480* (1.88) -0.093 (-0.10) .Latin America -1.575* (-1.90) -3.508*** (-3.26) .N 6720 6720 6720

25

error. Therefore, I calculated average annual TFP-growth rates between time-points where

Baier, Dwyer and Tamura (2006) provide TFP-estimates. Some of these periods are as short

as four years, and I include periods of up to twenty years. However, the large majority of

periods are ten years in duration (581 of 795). One period counts as one observation. For

the control variables, I used values at the start of the time period. However, for political

regime, I calculated the average of the PI over the five years prior to the period and all years

within the period except the five latest years, to account for the effect-lag on TFP-growth.

Democracy enhances TFP-growth also in these regressions. Independent of whether we

control for country-fixed effects or not, the effect from democracy is statistically significant

at least at the 5%-level. To sum up, there is robust evidence that democracy enhances

TFP-growth, as expected from the model above. Since TFP-growth is the main source of

long-term economic growth rates, there is also reason to expect that democracy enhances

GDP per capita growth. However, we still need to investigate whether democracy improves

economic growth empirically

5.2 Economic growth

Due to Maddison’s and Polity’s extensive data sets, this study draws on data from 154 exist-

ing and previous countries: the number of observations is around 8800 for models using 5-year

lags on independent variables, which is for example about twice the number of observations

used in (Przeworski et al. 2000). Models using 2-year lags draw on around 9400 observations.

Table 2 shows a positive effect from democracy on economic growth for the models using

5-year lags. The PCSE and RE models show a significant effect at least at the 5% level. This

is true also for models using a 2-year time lag, but the effect is not robust when using 3-year

lags. The estimated effect of going from least to most democratic according to the 2-year-

and 5-year lag PCSE and RE models, vary between 0.6 and 1.0 percentage points extra

annual GDP per capita growth. The FE-models, however, do not show significant effects,

26

Table 2: Results from OLS PCSE, RE and FE models with 5-year lags and GDP per capitagrowth as dependent. Time dummies and constant omitted from table. *** p < 0.01, **p < 0.05, * p < 0.10

Variable—Model OLS PCSE RE FEb (t) b (t) b (t)

Polity 0.036*** (2.65) 0.032** (2.26) 0.020 (1.28)Ln GDP pc -0.631*** (-2.67) -1.621*** (-8.80) -3.181*** (-13.05)Ln Population -0.005 (-0.09) -0.311*** (-3.33) -1.637*** (-6.29)Ln Regime dur. -0.038 (-0.62) -0.042 (-0.68) 0.002 (0.04)Ethnic fraction. -0.604 (-1.63) -0.753 (-1.12) .British 0.210 (0.78) 0.105 (0.25) .French -0.550* (-1.73) -1.056** (-2.04) .Portuguese 1.174** (2.19) 0.184 (0.19) .Spanish 1.591*** (2.88) 1.626** (1.99) .Belgian -0.030 (-0.03) -1.377 (-1.05) .Sunni -0.942 (-1.53) -1.388 (-0.86) .Shia -3.046*** (-3.12) -2.990* (-1.72) .Catholic -0.279 (-0.34) -0.274 (-0.16) .Protestant+ -0.529 (-0.70) -0.143 (-0.08) .Orthodox -0.942 (-1.05) -1.203 (-0.68) .Hindu -0.395 (-0.44) 0.351 (0.19) .Buddhist+ 1.238 (1.44) 0.882 (0.49) .Indigenous -1.667** (-2.46) -1.235 (-0.70) .E.Eur.+ Soviet rep. 0.289 (0.78) 0.147 (0.23) .Africa SS -1.720*** (-2.94) -3.817*** (-5.20) .Asia-Pacific -1.430* (-1.70) -2.593*** (-2.83) .MENA 0.570 (1.01) -0.048 (-0.06) .Latin America -2.429*** (-4.24) -3.657*** (-4.37) .N 8822 8822 8822

but the estimated effect is positive. Nevertheless, these models utilize a stringent inference

procedure as they do not allow cross-country comparisons to affect results. Moreover, when

we substitute the decade dummies with dummies that capture broader time period effects

(1871-1913, 1914-1945, 1946-1972, 1973-2003), the FE-results are also significant at least at

the 5%-level for models using 2-, 3- and 5-year lags.

The results above are in line with some of the more recent studies on democracy and

growth (e.g. Tavares andWacziarg 2001; Baum and Lake 2003; Bueno de Mesquita et al. 2003;

27

Papaioannou and Siourounis 2008; Feng 2005), including the very thorough meta-study by

Doucouliagos and Ulubasoglu (2008). The latter study shows that the sample’s time period

influences results, and that studies drawing on longer time series are more reliable. This is

good news for this study, as it draws on the longest time series in the literature. Although

the different models above build on different assumptions, we most often find statistically

significant effects at least at the 5%-level, and all models show a positive estimated effect.

But, how robust are the results? First, I relaxed the assumption that democracy’s ef-

fect on growth is linear. I used the nnmatch command in STATA to estimate the average

treatment effect (ATE) from democracy on growth with non-parametric, nearest neighbor

matching. Replacement is allowed in the estimation procedure. I used robust standard

errors (see Abadie and Imbens 2002).12 The unit of analysis is country-year. In practice,

we therefore draw on variation stemming from a single country changing regime type, and

on information from comparisons between relatively similar countries with different regime

types. The Polity-variable is dichotomized, and I chose two different cut-off points for democ-

racy (≥ 0 and ≥ 4) to check the results’ robustness. I also explored different numbers of

matches (one, three, five and ten closest matches). All democracy-coefficients are positive

and significant at the 5%-level, and almost all are significant at the 1%-level.13 The ATE

is positive and substantial in size. Some of the matching-results indicate a far larger effect

from democracy than the linear models; sometimes democracy yields an ATE larger than

2% extra GDP per capita growth.

Second, I investigated whether endogeneity of regime type might drive the results above.

One simple way to investigate endogeneity is through a simple Granger-test. Democracy

in t − 1 was estimated to have a positive, significant effect (1%-level) on growth in t, even

when controlling for growth in t − 1. When democracy in t was entered as a function of

12The calculation of these standard errors are based on 1 match with relatively similar units having similarvalues on the treatment variable.

13These models use a 2-year time lag.

28

democracy in t − 1 and growth in t − 1, the estimated effect from growth was actually

negative and significant at the 5%-level. According to this analysis, one should actually

expect that the models above underestimate the positive effect from democracy on economic

growth. However, the effect from democracy on growth is not robust when we run panel-

data 2SLS analysis (G2SLS), using lagged polity-values (see Helliwell 1994) and/or WAVE, a

dummy based on whether the current regime change originated in one of Huntington’s reverse

waves (see Knutsen 2011), as instruments.14 The G2SLS coefficients are mostly insignificant.

However, Hausman tests did not reject the hypothesis of similar coefficients, even at the 10%-

level, when comparing coefficients from (efficient) RE and (consistent) G2SLS models. One

interpretation of this result is that the endogeneity problem is not severe, and that we can use

the efficient RE-results; these show a statistically significant, positive effect from democracy.

The results can thus be interpreted to support the hypothesis that democracy enhances

growth. However, the results above are not completely robust. Empirically, there is large

variation in growth rates among democracies (e.g. Persson and Tabellini 2003) and especially

dictatorships (Rodrik 2000; Przeworski et al. 2000; Besley and Kudamatsu 2007; Knutsen

2011). Thus, degree of democracy does not, by far, explain all variation in economic growth

rates. As a large literature has documented, other institutional structures also matter for

growth. Moreover, the effect from democracy and dictatorship on growth may be contingent

on other factors, such as a leader’s personal characteristics, existence of alternative institu-

tional or non-institutional checks on rulers, and type of security threat facing a regime (see

e.g. Jones and Olken 2005; Wintrobe 1998; Olson 1993; Przeworski et al. 2000; Bueno de

Mesquita et al. 2003; Besley and Kudamatsu 2007; Acemoglu and Robinson 2006a; Doner,

Ritchie and Slater 2005; Knutsen 2011). For example, empirical results clearly indicate the

effect from democracy on growth is significantly more positive among countries with weak

14Country-years where the reigning regime originated in the periods ⟨←, 1827], [1922, 1942], [1958, 1975],[1998, 2003] are scored a 1 on WAVE.

29

state capacity (Knutsen 2009).

Moreover, the effect from democracy on growth may have changed over time. The main

sources of economic growth may increasingly be knowledge-related factors (see e.g. Galor

and Moav 2004; Goldin and Katz 2001; Houghton and Sheehan 2000). If economic growth

increasingly is determined by human capital and technology diffusion, and decreasingly by

physical capital, democracy’s growth advantage may have been larger in 1995 than in 1895.

Moreover, the increasing volume of FDI could mean that domestic savings rates, the ”com-

parative advantage” of dictatorships, is less important even for physical capital accumula-

tion. Empirical studies indicate that democracy may increase incoming FDI-volumes (see

e.g. Busse and Hefeker 2007; Li and Resnick 2003). I tested whether democracy had a more

positive effect on growth over the last decades, by conducting Chow-tests (see e.g. Greene

2003). I added an interaction term, where PI is multiplied with a post-1980 dummy, to the

OLS PCSE, RE and FE models above. There is indeed quite robust evidence that the effect

from democracy has become more positive after 1980. Both when using two- and five-year

lags on the independent variables, the interaction term is always positive and significant at

least at the 10%-level.15 The only exception is the OLS PCSE model, using a five-year time

lag. Here, the linear PI is significant at the 5%-level, but the interaction term is insignificant

at conventional levels.

6 Conclusion

The statistical analyses conducted in this article, based on the most extensive data-sample

in the literature, indicate that democracy has a positive impact on economic growth. The

effect from democracy on growth is relatively, but not completely, robust. The results above

reinforce the results in for example Baum and Lake (2003), Bueno de Mesquita et al. (2003),

15The results do not change much when we use a post-1985 dummy rather than a post-1980 dummy.

30

Feng (2005), and Doucouliagos and Ulubasoglu (2008), which have recently challenged the

widespread ”agnostic position” on democracy and growth. These authors correctly point

out that increased human capital accumulation and strengthened property rights protection

are two of the main reasons why democracy enhances growth. However, the theoretical

model presented above indicate an additional important reason: In dictatorships, technology

diffusion is slowed down because dictators manipulate civil liberties and promote policies

that inhibit idea diffusion. Although dictators in an optimal world may have wanted to

promote technological change to increase their own consumption, dictators in practice restrict

technological change as they are unable to perfectly separate politically dangerous from

economic efficiency-enhancing information. Empirical results, based on a very extensive

data material, corroborated this hypothesis; democracies have higher TFP-growth rates.

In the long run, therefore, democracies should prosper more than dictatorships. Physical

capital accumulation, where dictatorships often perform well (Tavares and Wacziarg 2001;

Przeworski and Limongi 1993), probably only has a transitional effect on growth (Solow

1956; Barro and Sala-i Martin 2004). The effect from technological change on growth is not

transitional (Romer 1990; Barro and Sala-i Martin 2004). Moreover, technology diffusion,

and human capital accumulation, may have become more important for growth over the

last decades, and this should increase democracy’s growth advantage over dictatorship. This

paper provides evidence also for this hypothesis; democracy’s effect on growth was higher

after 1980 than before.

31

References

Abadie, Alberto and Guido W. Imbens. 2002. “Simple and Bias-Corrected Matching Estima-

tors for Average Treatment Effects.” Washington D.C.: NBER. NBER Technical Working

Paper NO 283.

Abramowitz, Moses. 1956. “Resource and Output Trends in the United States since 1870.”

American Economic Review 46:5–23.

Acemoglu, Daron and James A. Robinson. 2006a. “Economic Backwardness in Political

Perspective.” American Political Science Review 100(1):115–131.

Acemoglu, Daron and James A. Robinson. 2006b. Economic Origins of Dictatorship and

Democracy. New York: Cambridge University Press.

Acemoglu, Daron, Simon Johnson and James A. Robinson. 2001. “The Colonial Origins

of Comparative Development: An Empirical Investigation.” American Economic Review

91(5):1369–1401.

Adsera, Alicia, Carles Boix and Mark Payne. 2003. “Are You Being Served?; Political

Accountability and Quality of Government.” RES Working Paper 4241, Inter-American

Development Bank, Research Department, Washington D.C.

Aghion, Phillipe and Peter Howitt. 1992. “A Model of Growth through Creative Destruc-

tion.” Econometrica 60:323–351.

Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat and Romain

Wacziarg. 2003. “Fractionalization.” Journal of Economic Growth 8(June):155–194.

Baier, Scott L., Gerald P. Dwyer Jr. and Robert Tamura. 2002. “How Important Are Capital

and Total Factor Productivity for Economi Growth?” Federal Reserve Bank of Atlanta.

Working Paper 2002-2a.

32

Baier, Scott L., Gerald P. Dwyer and Robert Tamura. 2006. “How Important are Capital

and Total Factor Productivity for Economic Growth?” Economic Inquiry 44(1):23–49.

Barro, Robert J. and Xavier Sala-i Martin. 2004. Economic Growth. Cambridge, MA.: The

MIT Press.

Baum, Matthew A. and David A. Lake. 2003. “The Political Economy of Growth: Democracy

and Human Capital.” American Journal of Political Science 47(2):333–347.

Beck, Nathaniel and Jonathan N. Katz. 1995. “What to Do (and Not to Do) with Time-Series

Cross-Section Data.” American Political Science Review 89(3):634–647.

Beetham, David. 1999. Democracy and Human Rights. Cambridge: Polity Press.

Besley, Timothy and Masayuki Kudamatsu. 2007. “Making Autocracy Work.” Working

Paper.

Boix, Carles. 2003. Democracy and Redistribution. Cambridge: Cambridge University Press.

Bueno de Mesquita, Bruce, Allistair Smith, Randolph M. Siverson and James D. and Morrow.

2003. The Logic of Political Survival. Cambridge, Massachusetts: MIT Press.

Burkhart, Ross E. and Michael S. Lewis-Beck. 1994. “Comparative Democracy: The Eco-

nomic Development Thesis.” American Political Science Review 88(4):903–910.

Busse, Matthias and Carsten Hefeker. 2007. “Political risk, institutions and foreign direct

investment.” European Journal of Political Economy 23(2):397 – 415.

Chang, Ha-Joon. 2006. The East Asian Development Experience The Miracle, the Crisis

and the Future. London: Zed Books.

33

Clague, Christopher, Philip Keefer, Stephen Knack and Mancur Olson. 2003. Property and

Contract Rights in Autocracies and Democracies. In Democracy, Governance and Growth,

ed. Stephen Knack. Ann Arbor: The University of Michigan Press pp. 78–109.

Denison, Edward F. 1962. The Sources of Economic Growth in the US and the Alternatives

before Us. New York: Committee For Economic Development.

Denison, Edward F. 1968. Why Growt Rates Differ: Postwar Experiences in Nine Western

Countries. Washington DC: Brookings Institution.

Diamond, Larry. 2008. The Spirit of Democracy. New York: Times Books.

Doner, Richard F., Bryan K. Ritchie and Dan Slater. 2005. “Systemic Vulnerability and the

Origins of Developmental States: Northeast and Southeast Asia in Comparative Perspec-

tive.” International Organization 59(2):327–361.

Doucouliagos, Hristos and Mehmet Ali Ulubasoglu. 2008. “Democracy and Economic

Growth: A Meta-Analysis.” American Journal of Political Science 52(1):61–83.

Easterly, William. 2001. The Elusive Quest for Growth: Economists’ Adventures and Mis-

adventures in the Tropics. Cambridge, MA: MIT Press.

Engerman, Stanley L., Elisa Mariscal and Kenneth L. Sokoloff. 1998. “Schooling, Suffrage

and the Persistence of Inequality in the Americas, 1800-1945.” Los Angeles: UCLA, De-

partment of Economics. Working Paper.

Evans, Peter. 1995. Embedded Autonomy. States & Industrial Transformation. Princeton:

Princeton University Press.

Feng, Yi. 2005. Democracy, Governance and Economic Performance. Cambridge, MA.: The

MIT Press.

34

Galor, Oded and Omer 2004 Moav. 2004. “From Physical to Human Capital Accumulation:

Inequality and the Process of Development.” Review of Economic Studies 71(4):1001–1026.

Goldin, Claudia and Lawrence F. Katz. 2001. “The Legacy of U.S. Educational Leadership:

Notes on Distribution and Economic Growth in the 20th Century.” American Economic

Review 91(1):18–23.

Greene, William H. 2003. Econometric Analysis. 5th Edition. Upper Saddle River: Prentice-

Hall.

Grossman, Gene M. and Elhanan Helpman. 1991. “Quality Ladders in the Theory of

Growth.” The Review of Economic Studies 58(1):43–61.

Hachigian, Nina. 2002. “The Internet and Power in One-Party East Asian States.” The

Washington Quarterly 25(3):41–58.

Halperin, Morton H., Joseph T Siegle and Michael M. Weinstein. 2005. The Democracy

Advantage: How Democracies Promote Prosperity and Peace. New York: Routledge.

Helliwell, John F. 1994. “Empirical Linkages between Democracy and Economic Growth.”

British Journal of Political Science 24(2):225–248.

Helpman, Elhanan. 2004. The Mystery of Economic Growth. Cambridge, MA: The Belknapp

Press.

Houghton, John and Peter Sheehan. 2000. “A Primer on the Knowledge Economy.” Center

for Strategic Economic Studies, Victoria University, Melbourne. Paper.

Jones, Benjamin F. and Benjamin A. Olken. 2005. “Do Leaders Matter? National Leadership

and Growth since World War II.” The Quarterly Journal of Economics 120(3):835–864.

35

Kihl, Young Whan and Hong Nack Kim, eds. 2006. North Korea: The Politics of Regime

Survival. New York: M.E.Sharpe.

Klenow, Peter J. and Andres Rodriguez-Clare. 1997. The Neo-Classical Revival in Growth

Economics: Has it Gone to Far? In NBER Macroeconomics Annual 1997, ed. Ben S.

Bernanke and Julio J. Rotemberg. Cambridge, MA: MIT Press.

Knack, Steven and Philip Keefer. 1995. “Institutions and Economic Performance: Cross-

Country Tests Using Alternative Measures.” Economics and Politics 7:207–227.

Knutsen, Carl Henrik. 2007. Democracy and property rights. Master’s thesis University of

Oslo, Department of Economics Oslo: .

Knutsen, Carl Henrik. 2009. “The African Growth Tragedy Revisited: Strong States, Weak

Rulers.” GARNET Working Paper No: 71/09.

Knutsen, Carl Henrik. 2010a. “Democracy, Dictatorship and Protection of Property Rights.”

Oslo: University of Oslo, Department of Political Science. Working Paper.

Knutsen, Carl Henrik. 2010b. “Investigating the Lee-Thesis: How Bad is Democracy for

Asian Economies?” European Political Science Review Forthcoming.

Knutsen, Carl Henrik. 2011. The Economic Effects of Democracy and Dictatorship PhD

thesis Department of Political Science, University of Oslo Oslo: . Forthcoming.

Kremer, Michael. 1993. “The O-Ring Theory of Economic Development.” The Quarterly

Journal of Economics 108(3):551–575.

Landes, David. 2003. The Wealth and Poverty of Nations. London: Abacus.

Leblang, David A. 1996. “Property Rights, Democracy and Economic Growth.” Political

Research Quarterly 49(1):5–26.

36

Leblang, David A. 1997. “Political Democracy and Economic Growth: Pooled Cross-

Sectional and Times-Series Evidence.” British Journal of Political Science 27(3):453–466.

Leftwich, Adrian. 2000. States of Development: On the Primacy of Politics in Development.

Cambridge: Polity Press.

Li, Quan and Adam Resnick. 2003. “Reversal of Fortunes: Democratic Institutions and

Foreign Direct Investment Inflows to Developing Countries.” International Organization

57(1):175–211.

Lindert, Peter H. 2005. Growing Public: Social Spending and Economic Growth since the

Eighteenth Century. Volume 1. Cambridge: Cambridge University Press.

Maddison, Angus. 2006. The World Economy. Paris: OECD.

Maddison, Angus. 2007. Contours of the World Economy, 1-2030 AD. Oxford: Oxford

University Press.

Mankiw, Gregory N., David Romer and David N. Weil. 1992. “A Contribution to the

Empirics of Economic Growth.” The Quarterly Journal of Economics 107(2):407–437.

Marshall, Monthy G. and Kaggers Keith. 2002. “Polity IV Project Dataset Users Manual.”

College Park: Program Center for International Development and Conflict Management,

University of Maryland. Manual.

Mauro, Paolo. 1995. “Corruption and Growth.” The Quarterly Journal of Economics

110(3):681–712.

Meltzer, Allan and Scott Richards. 1981. “A rational theory of the size of government.”

Journal of Political Economy 89(5):914–927.

Mill, John Stuart. 1974. On Liberty. London: Penguin.

37

Mueller, Dennis C. 2003. Public Choice III. Cambridge: Cambridge University Press.

Murphey, Rhoads. 2000. A History of Asia. New York: Addison-Wesley-Longman.

Nelson, Richard R. 2005. Technology, Institutions and Economic Growth. Cambridge, MA:

Harvard University Press.

Nelson, Richard and Sidney Winter. 1982. An Evolutionary Theory of Economic Growth.

Cambridge, MA: Harvard University Press.

North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance.

Cambridge: Cambridge University Press.

North, Douglass C. 2005. Understanding the Process of Economic Change. Princeton: Prince-

ton University Press.

North, Douglass C., John Joseph Wallis and Barry R. Weingast. 2009. Violence and Social

Orders: A Conceptual Framework for Interpreting Recorded Human History. Cambridge:

Cambridge University Press.

Olson, Mancur. 1982. The Rise and Decline of Nations. Economic Growth, Stagflation and

Social Rigidities. New Haven: Yale University Press.

Olson, Mancur. 1993. “Democracy, Dictatorship and Development.” The American Political

Science Review 87(3):567–576.

Overland, Jody, Kenneth L. Simmons and Michael Spagat. 2000. “Political Instability and

Growth in Dictatorships.” London: Royal Holloway College, University of London. Work-

ing Paper Number 354.

Papaioannou, Elias and Gregorios Siourounis. 2008. “Democratization and Growth.” De-

partment of Economics, University of Peloponnese: Working Paper 2008-27.

38

Persson, Torsten and Guido Tabellini. 2003. The Economic Effects of Constitutions. Cam-

bridge, Massachusetts: Cambridge, MA: MIT Press.

Pinto, Pablo M. and Jeffrey. F. Timmons. 2005. “The Political Determinants of Economic

Performance: Political Competition and the Sources of Growth.” Comparative Political

Studies 38(1):26–50.

Przeworski, Adam and Fernando Limongi. 1993. “Political Regimes and Economic Growth.”

Journal of Economic Perspectives 7(1):51–69.

Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub and Fernando Limongi. 2000.

Democracy and Development. Political Institutions and Well-Being in the World, 1950-

1990. Cambridge: Cambridge University Press.

Robinson, James A. 1998. “Theories of ”Bad Policy”.” Policy Reform 2(1):1–46.

Robinson, James A. 2001. “When is A State Predatory.” Cambridge, MA.: Harvard Uni-

versity. Working Paper.

Rock, Michael. 2009. “Corruption and Democracy.” The Journal of Development Studies

45(1):55–75.

Rodrik, Dani. 1997. “TFGP Controversies, Institutions, and Economic Performance in East

Asia.” Cambridge, MA.: NBER Working Paper No. 5914.

Rodrik, Dani. 1999. The New Global Economy and Developing Countries: Making Openness

Work. Baltimore: The Johns Hopkins University Press.

Rodrik, Daniel. 2000. “Institutions for high-quality growth: What they are and how to

acquire them.” Studies in Comparative International Development 35(3):3–31.

39

Romer, Paul. 1990. “Endogenous Technological Change.” Journal of Political Economy

98(5):71–102.

Romer, Paul. 1993. “Idea Gaps and Object Gaps in Economic Development.” Journal of

Monetary Economics 32(3):543–573.

Sen, Amartya. 1999. Development as Freedom. New York: Anchor Books.

Solow, Robert M. 1956. “A Contribution to the Theory of Economic Growth.” Quarterly

Journal of Economics 70:65–94.

Solow, Robert M. 1957. “Technical Change and the Aggregate Production Function.” Review

of Economics and Statistics 39:214–231.

Stasavage, David. 2005. “Democracy and education spending in Africa.” American Journal

of Political Science 49(2):343–358.

Tavares, Jose and Romain Wacziarg. 2001. “How democracy affects growth.” European

Economic Review 45(8):1341–1378.

Tsebelis, George. 2002. Veto Players How Political Institutions Work. Princeton: Princeton

University Press.

Verspagen, Bart. 2005. Innovation and Economic Growth. Vol. The Oxfors Handbook of

Innovation Oxford: Oxford University Press chapter 18, pp. 487–513.

von Hayek, Friedrich A. 1945. “The Use of Knowledge in Society.” The American Economic

Review 35(4):519–530.

Wade, Robert. 1990. Governing the Market: Economic Theory and the Role of Government

in East Asian Industrialization. Princeton University Press: Princeton.

40

Wintrobe, Ronald. 1990. “The Tinpot and the Totalitarian: An Economic Theory of Dicta-

torship.” The American Political Science Review 84(3):849– 872.

Wintrobe, Ronald. 1998. The Political Economy of Dictatorship. Cambridge: Cambridge

University Press.

41