The Economic Effects of Democracy an Empirical Analysis *

The Economic Effects of Democracy – An Empirical Analysis[*]

Carl Henrik Knutsen

Draft Working Paper, Political Science Department, University of Oslo, March2008

Abstract

There is no existing consensus among academics on how democracy as a political regime type affects economic growth. There are many different theoretical arguments,sketching up potential causal mechanisms, pointing both in favor and disfavor of democracy when it comes to its ability to generate economic growth. Also when it comes to empirical research, results have been diverging, and this goes for case studies and small-n comparative studies, as well as statistical studies on the possible relationship. This is a purely empirical paper, and its aim is to investigate whether there is enough reasonable evidence to claim that democracy as a regime type has any systematic influence on national growth rates in GDP per capita. In addition, the effects of democracy on protection of property rights and corruption are also investigated, although in a less thorough manner.

The paper contributes to the already large existing body of empirical researchon the topic by using a very extensive data set with the newest available data, and particularly by utilizing different, and more appropriate, statistical techniques when compared to a large bulk of the existing studies. This study, using country-years as the unit of analysis, investigates the possible relationship between democracy and economic factors by applying OLS with Panel Corrected Standard Errors, Fixed-Effects, Random-Effects, 2SLS and Matching-based techniques to the data. The study also describes some often overlooked methodological problems related to this issue.

According to the analyses in this paper, there is good reason to believe that democracy “on average” increases economic growth, increases protection of property rights and reduces corruption. Nevertheless, the results are not unequivocal, as the results vary withthe specified model, the choice of democracy indicator as well as the statistical technique applied.

1.Introduction

“This is a fairly boring and tedious paper, and is intentionally so. This paper provides no new interpretations of the East Asian experience to interest the historian, derives no new theoretical implications of the forces behind the East Asian growth process to motivate the theorist, and draws no new policy implications from the subtleties of East Asian government intervention to excite the policy activist.” (Young, 1995:641)

The quote above is taken from Alwyn Young’sarticle “The Tyranny of Numbers”, analyzing the sources of growth in four East Asian nations. If one substitutes “East Asian” with “global”, the quote fairly precisely describes the paper you are about to read as well. This paper is a purely empirical one, seeking to investigate whether democracy as a regime type over the last decades “on average” has influenced national economic growth rates and other economic factors in either a positive or negative direction. Democracy and economic development are more often than not viewed as inherent normative goods. Therefore, the question on whether these concepts are empirically related is an important question to address. Do we generally face a trade-off between two goods, or do the two go together? I will here ask the “whether/if”- and “how” questions, aiming at general description of aggregate relationships, and leave the “why” question to another day.

Does democracy systematically affect aspects of an economy, like economic growth rates? The answers to this question have come in at leastfour forms: “Yes, positively”, “Yes, negatively”, “No” or “We don’t know” (see Przeworski and Limongi (1993) and Knutsen (2006a:229-235) for literature overviews). The last two answers, that there is no general and strong effect from democracy on economic growth, or that it has at least not been detected, have been the prevailing views among scholars after Przeworski et al.’s seminal and methodologically thorough study from 2000[1].Nevertheless, the lack of consensus on the issue and the divergence in estimated results between papers warrants a new statistical investigation into this general and important topic. What can this paper possibly contribute with to the already large volume of research? First, it provides econometric models that I believe are well specified; it controls for relevant prior causal variables, and leaves out variables that should be excluded. Second, it specifies some methodological problems connected to the research question, which some but not all researchers seem to be aware of. Third it utilizes new data, and thereby extends the sample period when compared with earlier research. Fourth, and I believe this is the most important justification, it uses“new”, and relatively advanced statistical techniques, which might be better suited than for example OLS to deal with the question of interest.

Let it be clear from the outset: Empirical results not followed by specified theories, models or mechanisms provide limited insight. Additionally, the relationship between democracy and economic growth is most certainly dependent on context and also on the type of democracy and authoritarian regime we are comparing, and average effects do not exhaust the complex nature of the relationship. Nevertheless, the general relationship is still of interest. People will not stop asking about general relationships even if there exist no deterministic laws of Newtonian character (even Newton’s laws are modified in certain contexts, like Einstein showed us). Relationships in the social sciences are nearly always probabilistic and “full of exceptions”. This being said; rather than claiming that regression-coefficients exhaust all potential information of the relationship, one should recognize that there are multiple plausible causal mechanisms that could be at work, and that the relative strength of these mechanisms depend on several contextual factors connected to the economic, political, social and maybe even cultural environment. I have earlier evaluated no less than 20 arguments, identifying 20 potential causal mechanisms with the degree of democracy in a country as a cause and the rate of growth in GDP per capita as an effect (Knutsen, 2006a), and analogously five arguments when protection of property rights is the effect variable (Knutsen, 2007). The arguments point in different directions when it comes to the effects of democracy, and some of the arguments seem to be more valid in specific situations (both spatially and temporally). This does not however preclude us from asking whether we can identify any overall pattern in the data. Do certain causal mechanisms dominate in the real world and thereby influence the empirically observed relationship in a systematic way? “Does a higher degree of democracy produce more economic growth in general?” is maybe an imprecise question, but we can still seek to provide some evidence for a rough sketch of an answer.

1.1 Less than one page on the “why” question

When it comes to democracy and economic growth, two arguments are central in pointing out the deficiencies of democracy. The first states that in democracies, the electorate will press demands for private and public consumption that democratically elected leaders, as opposed to authoritarian rulers, cannot afford to ignore. Increased consumption implies less saving and thereby less investment and ultimately less economic growth (Solow, 1956). The second argument is that because of certain political mechanisms, it will be harder for democratic leaders than more autonomous authoritarian leaders to push through contested reforms that are eventually economic growth-enhancing. On the other hand, democracies are argued to increase economic growth through among others facilitating innovation and diffusion of new technologies, which are crucial to long term growth rates (Romer, 1993 and Fagerberg, 2002). Democracy might also increase the legitimacy of economic reforms and also enhance growth through providing political checks on leaders that would otherwise have acted in ways that would have benefittedonly themselves or narrow backing groups economically, at the cost of wider society (Knutsen 2006a). When it comes to the protection of property rights, democracy is also claimed to have a positive influence on this variable through the same political checks on elite-led grabbing of property.On the other hand, democracy might have a negative influence on the stability of existing property rights arrangements because a relatively poor majority will have incentives to redistribute from richer minorities (Knutsen, 2007).The effect from democracy on overall corruption in a society is often assumed to be negative because of the existence of several accountability generating mechanisms like elections and freedom of press. However,democracy might decentralize corruption (Knutsen 2006a:114-115), and decentralized corruption is argued to be more harmful to economic growth than centralized corruption (Shleifer and Vishny, 1993). These are only a subset of the existing arguments on possible causal effects from degree of democracy on different aspects of the economy. Democracy is also argued to affect other aspects of the economy like income inequality and “human capital”, but these issues will not be addressed here.

1.2 Outline of the paper

Section 2 of this paper deals with the conceptual definition and operationalization of the independent variable, political regime type, and the different dependent (economic) variables analyzed in this paper. Section 3 deals with a cluster of identified methodological problems related to testinghypothesesin this issue-area. Section 4 contains the description of the different methodologies and techniques used, and the empirical analysis of how democracy affects economic growth. Section 5 gives a brief overview overempirical results on how democracy affects protection of property rights and corruption. Section 6 draws up the main conclusions.

2.Definitions and operationalizations

2.1 Democracy

There is no consensus on how to define democracy conceptually. Crick (2002:1) calls it “an essentially contested concept”. I have earlier presented and debated different ways of understanding and defining democracy conceptually, and I will not go deep into the question here (Knutsen, 2006a, 2006b and 2006c). Two categories of definitions are one the one hand those defining democracy according to a particular set of institutions and on the other hand the definitions that define democracy according to some underlying principles. The latter type of definition is sometimes referred to as “substantial” definitions of democracy. One such particular definition is the one proposed by David Beetham (1999), and I have earlier argued in favor of understanding democracy as a concept along the lines Beetham sketches up. Beetham proposes that “[T]he core idea of democracy is that of popular rule or popular control over collective decision making” (Beetham, 1999:90). Authoritarian governments are therefore related to the lack of popular control over collective decision making, since a low degree of democracy implies authoritarianism. To this first criterion, another one of political equality among citizens is added by Beetham. Democracy “is realized to the extent that such [public] decision-making actually is subject to the control of all members of the collectivity considered as equals” (Beetham, 1994:28). These two principles are the “first level” of the (democracy) concept; the deeper principles (Goertz, 2005).Then we have the “second level” of the concept (Goertz, 2005). This level points to the empirical features that help realize the underlying principles of democracy. Institutional structures like for example competitive elections of legislators are crucial elements.Beetham also argues that guaranteed political rights and civil liberties are necessary prerequisites for democracy. “The freedoms of speech, association, assembly and movement, the right to due legal process, and so on, are not something specific to a particular form of democracy called ‘liberal democracy’; they are essential to democracy as such, since without them no effective popular control over government is possible” (Beetham, 1994:29). The third level of the democracy concept, the indicator-level, relates to the practical operationalization of democracy in terms of measurable characteristics, and the choice of operationalization is discussed below.

When it comes to institutionally based definitions, these have some obvious beneficial properties in terms of precision, and they can more easily be operationalized in stringent ways. Przeworski et al.’s (2000) definition of democracy provides a good example of such a definition. Another famous institutional definition of democracy is Schumpeter’s definition of “the democratic method” as “that institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people’s vote”(Schumpeter,1976:269).However, as Beetham (1999) claims, if one is to dig deeper into institutional definitions of democracy, a pressing question is exactly why the particular proposed set of institutions makes up democracy. Any answer risks throwing us out into circular argumentation if we do not establish an underlying idea of what democracy actually is.

Since I am following Beetham’s substantial definition of democracy closely, it is clear that it is preferable to operationalize democracy in a relatively broad sense. This means that it does not suffice to look for existence of elections or degree of checks on the executive when I choose my favored operationalization. Existence of institutions clearly needs to be part of any operationalization, but in addition the actual functioning of these institutionsmust be included; that is, we are interested in whether these institutions actually enhance the two principles that serve as the foundation of the conceptual definition of democracy. The existence of multi-party elections and constitutional descriptions of freedom of speech and assembly are not sufficient for popular control over politics or political equality. We need subjective evaluations of the functioning of these institutional structures in practice, and we also need toincorporate the degree to which civil liberties and political rights are existent. I have therefore argued that the Freedom House Index (FHI), despite its many shortcomings, is the best available cross-country index, with a relatively extensive time series, for democracy (Knutsen 2006a, Knutsen 2006c). The FHI has been attacked, both because it does not sufficiently incorporate the actual functioning of institutional structures (Inglehart and Welzel, 2006), but also from the “opposite side” because of possiblesubjectivity bias and incorporation of policy outcome-related aspects (Strand, 2007). Strand (2007) also points out that there has been a change in the FHI coding practice over the course of time. Nevertheless, among the existing available democracy-indexes, I find the FHI to be the better alternative because it balances both the need to be subjective in order to capture “the political reality behind the formal and observable structures of government” (Grugel, 2002:22), and the problem of subjectivity bias and outcome-centeredness on the other hand.

The FHI consists of 25 main check questions[2], 10 on political rights and 15 on civil liberties, leading to two different indexes, both ranging from 1 to 7. In this study, the “FHI-index” refers to the (unweighted) average of these two indexes.I will also use two additional operationalizations, in order to check the robustness of my findings. The first is the Polity IV measure of democracy (see Marshall and Jaggers (2002) for a discussion of operationalization), which runs from -10 to 10. The Alvarez-Cheibub-Limongi-Przeworski (ACLP) data set’s dichotomous measure of democracy is the second alternative measure (see Przeworski et al. (2000) for a discussion on operationalization). Note, when interpreting estimates, that a high value on the FHI and ACLP-measure of regime type indicates “less” democratic, and that a high value on the Polity indicates “more” democratic. The FHI data used in the analyses run from 1972 to 2005, the Polity data used in this study are from 1960 to 2004, and the ACLP data used here go from 1960 to 2002.

2.2 Economic growth

Economic growth in a geographic area is defined as the increase in the aggregate level of production during a year. I operationalize the concept by using growth in PPP-adjusted GDP per capita, with data collected from the World Development Indicators (WDI). I use PPP-adjustment instead of exchange adjustment because my main underlying concern is welfare effects, and I therefore need to take into account the local price level. There are several methodological problems with using GDP per capita growth as a proxy for economic growth. Depreciation of capital is not calculated into the concept, and it does not take into account depletion of natural resources either (oil and mineral extraction taken from known reserves). The most problematic issue when it comes to using it on a global sample however, is that it does not take into account the size/growth of the informal economy, which is relatively large especially in some developing countries

2.3 Protection of property rights and corruption

“A right of ownership is associated with property that establishes the good as being "one's own thing" in relation to other individuals or groups, assuring the owner the right to dispense with the property in a manner he or she sees fit, whether to use or not use, exclude others from using, or to transfer ownership” ( is one way of defining property rights. Development- and institutional economists in particular have been focusing on the importance of a well-functioning system of private property rights for the economy. “Property rights are seen as a key institutional feature because they provide actors with proper incentives in economic life, and well-structured property rights therefore enhance efficient resource allocation in an economy.” (Knutsen, 2007:8). I use the World Governance Indicator’s “Rule of Law” dimension as a proxy for the protection of private property rights in an economy, since a large bulk of the indicators used to construct it are related to different factors affecting the protection of these rights. However, as the name indicates, there are also some indicators (among the 50 or so utilized in the construction of the index) relating to other aspects of rule of law in a country. The index is constructed through a principal component analysis, with a specific weighting procedure when it comes establishing the relative importance of the indicators in the index (Kaufman et al., 2007).