‘Free-Market Capitalism’ and Democracy in the Period of Democratic Recession: Investigating the Relationship in 141 Countries, 2006–2017

Abstract Since the mid-2000s, democratization has slowed, stopped, and even reversed across the world. At the same time, societies have become more oriented toward free markets as measured by indexes of economic freedom. Relying on a panel sample of 141 developed and developing countries between 2006 and 2017, this paper is the first to investigate whether the two phenomena are related by employing economic freedom data. It finds that there is no net-negative relationship between aggregate economic freedom and democracy in this time-period. Instead, mixed findings of both an overall positive and overall neutral (but not negative) association are uncovered in between-country and within-country analyses, respectively. In between-country analyses, using the disaggregated index shows that the legal system/property rights component drives most of the positive relationship between aggregate economic freedom and democracy in the developed world. The same between-country analyses in the developing world show that freedom of international trade is positively associated with democracy, while modest regulation has a negative relationship. However, additionally controlling for omitted variable bias using country-fixed effects, the paper does not find evidence for either a positive or negative relationship between subsequent changes in levels of economic freedom and democracy.


Introduction
In the past decade and a half, the world has entered a protracted democratic recession (Puddington 2007;Diamond 2008;Diamond 2015;Bermeo 2016).After the decade-long third democratization wave , democracy expansion around the world has come to a sudden stop and a partial reversal (Freedom House 2022; V-Dem 2022).The plausible reasons for this change are many, but one salient potential cause has been the entrenchment and further expansion of the practices and institutions of 'free-market capitalism' (Bruff 2014;Brown 2019).
Free-market institutions, such as liberalized international trade, decreases in the size of government, deregulation, and expansion of private property rights, have for instance been plausibly seen as exacerbating income and wealth inequality, thus giving disproportionate social power to the rich (Harvey 2005;Wright 2010).The rich can use such asymmetric power to dismantle democratic checks on (their) political power so as to enrich themselves even further (by, for instance, reducing taxes) and to gain even more power.Moreover, free-market reforms can cause economic dislocation due to increased domestic and foreign competition, or economic frustration due to rising inequalities (Wright 2010;Bremmer 2018;Fukuyama 2019), which in turn can cause embittered workers to rebel against the status quo and elect illiberal populist leaders, such as Donald Trump, Viktor Orb� an, and Jair Bolsonaro, who have been the main proximate causes of the democratic recession.Harvey (2005:36-37) has claimed that, in the course of free-market reforms, '[t]he boundary between state and corporate power has become more and more porous.What remains of representative democracy is overwhelmed, if not totally though legally corrupted by money power.'He notes that there is 'a growing lack of symmetry in the power relation between corporations and individuals,' which throws into sharp relief the system's 'antidemocratic tendencies' (Harvey 2005:79, 195).He charges that free-market capitalism 'in its pure form has always threatened to conjure up its own nemesis in varieties of authoritarian populism and nationalism' (Harvey 2005:81).As a result, '[t]he democratic deficit in nominally "democratic" countries such as the US is now enormous.'(Harvey 2005:205) Klein has also spoken of a general 'potential incompatibility' of free-market capitalist reforms and a democratic system (Klein 2007:541).She has claimed that 'while Freidman's economic model is capable of being partially imposed under democracy, authoritarian conditions are required for the implementation of its true vision.' (Klein 2007:77-78) Her book, she writes, explicitly challenges the 'central and most cherished claim,' namely that 'free markets go hand in hand with democracy.'(Klein 2007:19) Harvey and Klein are perhaps the most well-known contemporary critics of free markets, but they are far from alone in their negative verdict.For instance, Patnaik (2020) has similarly argued that '[f]ascism is neoliberal capitalism's "gift" to mankind in the period of its maturity … the only way of transcending the fascist presence is to transcend neoliberal capitalism.' So, too, have Bruff (2014), Plehwe et al. 2020, andBrown (2019) argued that after a certain point, free-market capitalism goes hand in hand with political authoritarianism, illiberal populism, and a general decline in democracy.As (Slobodian 2019) has bluntly titled one of his articles, in which he discusses Milton Friedman and indexes of economic freedom, 'Democracy doesn't matter to the defenders of "economic freedom"'.
In his critique of Klein (2007), Norberg (2008) pointed out more than a decade ago that even a superficial glance at the growing number of democratic regimes in the world between 1990 and the present (the year 2007 at that time) seems to suggest that freemarket reforms could not be as inimical to democracy as she and other critics suggest.He points out that as the world tended more toward free markets over that period (both according to Klein's own account and according to indexes of economic freedom), so too has democratization significantly expanded, not shrunk.As he put it, 'contrary to the implications of Klein's thesis … while markets have been opened, the world has simultaneously undergone a democratic revolution.Between 1990 and 2007 the number of electoral democracies increased from 76 to 121.' (Norberg 2008:15) A decade and a half later, however, Norberg's simple rebuttal no longer stands.To the contrary.A new simple fact seems now to corroborate the critics' democratic suspicion about free-market capitalism, and is the initial motivating fact for this paper.According to various measures of political and civil rights across the world, most famously the Freedom House scores, we have been witnessing a protracted democratic recession for quite some time now.After decades of increasing robustness of existing, and the appearance of new, democracies, the trend has stopped and even reversed somewhat.Today, the world is no more democratic than 5, 10, or even 15 years ago; according to some measures, there are fewer and less robust democracies today than 15 years ago.At the same time, the most widely used scholarly indicator of free-market capitalism, the Fraser Institute's economic freedom index (Gwartney, et al. 2021), has been increasing during this period.Between 2006 and 2019, which is a common dating of the democratic recession, the economic freedom of the world has continuously been increasing from 6.86 to 7.04 (on a 0-10 scale).Put simply, as the world has grown more neoliberal from 2006 onward, it has also stopped growing more democratic and even backtracked somewhat.
This paper strives to answer the question of whether or not the two phenomena are related to each other and what the precise valence of the relationship is.In the first section that follows the introduction, I perform a review of the literature associated with measuring free-market institutions through indices of economic freedom and the literature on the democratic recession.The section also presents additional plausible theoretical mechanisms connecting free-market institutions and the democratic recession, considering reasons for thinking the association should be either positive or negative.The second section presents the data and method for the empirical part of the study, focusing especially on the main independent and dependent variables of interests and various control variables.The third section present the results of different regressions models, using aggregate and disaggregated economic freedom scores performed on both the whole sample and two subsamples of countries (developed and developing).The fourth section discusses and interprets the results, presenting ways of reconciling the mixed nature of the findings and ending with a broader conclusion.

Quantitative research on the social correlates of free-market capitalism
A large recent survey of over 700 quantitative studies (Lawson, et al. 2022) has demonstrated that the Fraser Institute's index of economic freedom mostly correlates positively with various 'good' social outcomes (like more economic prosperity, human rights, and peace), while it only rarely correlates positively with 'bad' social outcomes, such as increased corruption or increased inequality.Mixed, null, or uncertain results are also common, with the majority of associations between economic freedom and inequality (and corruption) falling in this latter category.Overall, 'Just over half, 50.6% [of examined studies], found economic freedom was related to "positive" outcomes while only 4.6% found "negative" outcomes; 44.8% did not find a clear relationship … ' (Lawson, et al. 2022:iii).The most clear and robust finding is that economic freedom correlates positively with income and growth -more than two thirds of the 143 relevant studies from the survey found such an effect.
As will be discussed in detail in the methodological section, it is debatable to what extent the economic freedom index is the most appropriate proxy measure of freemarket capitalism.I would like to note for now, however, that many prominent researchers, including economists, sociologists, and political scientists, have recently used indexes of economic freedom in just such a way.For instance, McLean et al. (2019) have used the index in the International Journal of Sociology to investigate the relationship between 'neoliberalism and homicide'.The economist Christian Bjørnskov (2015) has similarly examined the connection between 'neoliberal policies and homicide rates' by relying on the index, and he has also used the index to investigate the 'Hayek-Friedman' hypothesis regarding the positive influence of a free economy on press freedom (Bjørnskov 2018).The political scientists Indra de Soysa and Krishna Vadlamannati have recently used the index to answer whether 'free-market capitalism' drives unequal access to health (de Soysa and Vadlamannati 2021a), and income and social inequalities (de Soysa and Vadlamannati 2021b).Okunlola et al. (2022) have used the index to answer the question, 'Does a free market system reduce conflict in Africa?'.
Even though the economic freedom index is a widely used scholarly ranking, there have been surprisingly few studies investigating its relationship with democracy, and none exploring it in relation to the ongoing democratic recession.Boudreaux and Holcombe (2017) investigated whether a higher amount of market-oriented economic institutions (proxied by Fraser's index of economic freedom) works protectively against democratic breakdown.They found that it does, even controlling for various economic and geographic confounds.If a democratic government is established in a country within the first quartile of economic freedom, it survives in 90% of cases after 20 years.Democracies in the bottom quartile of economic freedom have only a 10% chance of survival after 20 years, and even just after 4 years half of them already break down.Bjørnskov (2018) looked specifically at the impact of the Heritage Foundation economic freedom index on press freedom, a key component of any democracy, and after controlling for various political and economic confounds found a positive association.However, when disaggregating the index into its subcomponents he discovered that only increased market openness is strongly associated with press freedom -the amount of regulation and government size did not matter either way.
Note, importantly, that these studies mostly leave out the crucial recent period of the democratic recession (2006-present).Moreover, there is precious little additional research on the contemporary impact of economic freedom on democracy.Several studies measure the impact of democracy on economic freedom (see, for instance, de Haan and Sturm 2003;Lundstr€ om 2005), but even these mostly rely on older samples from before the democratic recession.
Aside from the positive findings of Boudreaux and Holcombe (2017) and Bjørnskov (2018), there are other suggestive, if indirect, pieces of evidence that free-market capitalism might not be inimical to democracy.Despite some earlier critiques (Przeworski 2000;Acemoglu et al. 2008), the recent empirical literature has mostly vindicated a qualified, contemporary version of modernization theory, which is the idea that overall economic development increases the probability of a country transitioning to democracy and staying democratic (Boix 2011;Treisman 2020;Rød, Knutsen, and Hegre 2020).This finding is important for the present paper because most studies of economic freedom find that it contributes to more development and higher growth rates, which could mean that -if anything -economic freedom has a positive indirect effect (through the conduit of spurring economic development) on democratic transitions and democratic durability.
Further research on the relationship between economic freedom and democracy is needed, especially so because there are plausible theoretical mechanisms on both sides.As pointed out already in the introduction, more economic freedom could either directly or indirectly undermine democracy, particularly through the inequality channel and the economic-displacement or smaller-welfare channel.Moreover, deregulation could help businesses amass even more economic resources, which would in turn enable them to gather more political power with which they could dismantle democratic checks and balances.
However, more economic freedom could also increase economic development and contribute positively to democracy through the modernization channel.Robust economic development can lead to the replacement of rural farmers with the industrial working class, the replacement of landed elites with a mobile capitalist elite, and a swelling of the middle class (Acemoglu and Robinson 2006;Usmani 2018).These are socialstructural changes that theoretically increase the probability of democratization by changing both the incentives key social actors have for or against democratization as well as their capacities for collective action.For instance, industrial workers have an easier shot at paralyzing the economy through strike activity than farmers, which they can then use as a bargaining chip with the political (or economic) elite in their quest for democratization (Usmani 2018).Moreover, a smaller government, more international competition, and less regulation could lead to a dispersal of concentrated power on the part of large political and economic actors (special interests), or reduce their rentseeking ability, which might further enable and embolden civil society to clamor for democracy or to protect against incursions against it (Friedman and Friedman 1982).

The democratic recession
Already since 2007, democracy scholars like Arch Puddington (2007) and Larry Diamond (2008) have been consistently sounding the alarm that the enormous late-20th and early-21st century democratic gains had suddenly stopped and soon started reversing.Writing in 2015, Diamond (2015) forcefully repeated that this sudden end to democratic flourishing was no temporary blip, again insisting that 'The world has been in a mild but protracted democratic recession since about 2006.'Some democracy researchers were not completely convinced.Levitsky and Way (2015) argued, for instance, that global democracy scores provided by various indexes remained either stable or slightly increased between mid-2000s and 2013.They also claimed that both the overall number and share of democracies in the world had evinced either stability or only a very modest decrease between mid-2000s and 2013.Extending the data up till 2015, Brownlee (2017) has shown that although the number of electoral democracies has decreased substantially between 2006 (123) and 2010 (115), this trend had reversed by 2015, when the world again counted 125 electoral democracies.Some of this disagreement is merely the result of differences in emphasis ('is it a recession or is it stagnation?').A small part of it stems from differences in measures and indexes.But there is no denying that, compared to the period before the mid-2000s, the overall and broad trends in global democratization have changed significantly.The acclaimed democracy tracker Freedom House (2022) reports a steady increase in the share of 'Free' countries -which is Freedom House's category that groups together democracies with the most political and civil rights -from 32% in the early 1980s to 46% in the mid-2000s.The share has then stagnated and fallen since the late 2000s to 44% and even lower in late 2010s.Several fledgling new democracies, such as Serbia and Hungary, have deteriorated to such an extent that they lost their 'Free' designation over the past 5 years.Serbia's score declined from 76 (on a 0-100 scale, 100 being most democratic) in 2006 to 66 in 2019.Hungary's score went from 93 in 2006 to 70 in 2019.Even long-standing democracies, such as the UK, have been losing points in recent years (from around 97 in late 2000s to around 94 in late 2010s).Some, like the US, have even witnessed large drops of 5-10 points over the past 15 years, hovering in the low 80s nowadays.This is not just an artifact of Freedom House's measurements and (demanding) methodology.The same pattern of historic increase in the late twentieth century and then stagnation and decline in the last 15 years appears elsewhere.According to the V-Dem Institute (V-Dem 2022), another highly reputable venue for democracy measurement, the share of all democracies -combining both liberal and electoral democraciesincreased between the early 1980s and mid-2000s from around 25% to more than 50%.Since then, the share remained roughly stagnant.Liberal democracies in particular, that is, political regimes with the highest amount and best protection of political and civil rights, have declined from around 23% at their peak to around 20% in the last few years according to V-Dem.The V-Dem global liberal democracy index (scale of 0-1) has also steadily increased from 0.25 in the early 1980s to a high of 0.41 in the mid-2000s.The increase then completely stopped (and even declined somewhat to 0.39) over the last 15 years.The era of democracy expansion is clearly over, at least for now, and it has been so for some time (Figure 1).
One of the most discussed proximate mechanisms for the democratic recession is the rise of illiberal populism.According to two prominent scholarly accounts (for both see Norris and Inglehart 2018), voters in the past few decades have been electing illiberal populists leaders, who then start dismantling democratic institutions when they reach office, either because of increasing economic insecurity (due to economic displacement or reduction in government welfare) and broad economic frustration (due to rising income and wealth inequality) or because of cultural grievances (due to cultural mixing and the decline of the dominance within societies of previous traditional cultures).In turn, both economic and cultural causes of illiberal populism in particular and the democratic recession in general are regularly argued to have happened because of economic and cultural globalization, i.e., increased trade liberalization, deregulation, increased labor and capital migration, and so on, which are synonyms for free-market capitalism.As Francis Fukuyama (2019:168) recently put it, 'I concur with the commonplace judgment that the rise of populism has been triggered by globalization and the consequent massive increase in inequality in many rich countries.'Currently, there is no overwhelming and clear consensus whether what Fukuyama terms the 'commonplace judgment' is correct.Some studies find support for some type of the economic-grievances account for the rise of populists and the weakening of democracy (Colantone and Stanig 2018), while others do not (Bergh and K€ arn€ a 2021).Some find confirmatory evidence only or primarily for the cultural-backlash account (Norris and Inglehart 2018), while others dispute it (Sch€ afer 2022).This paper is related to, and contributes to, the literature on the democratic recession specifically and to the more general literature on the social correlates of economic freedom.Most concretely, I aim to contribute to the relatively sparse quantitative research investigating the potential synergies and/or tensions between contemporary capitalism and democracy.I do so by empirically examining the critical hypothesis that free-market capitalist reforms (as proxied by F-EFI in my analysis) have been positively associated with the ongoing global dismantling of democracy (as proxied by the Freedom House freedom rankings in my analysis).The main idea is that if there is no such correlation, then the thesis that free-market capitalism has fueled the democratic recession (either through the proximate mechanism of illiberal populism or some other mechanism) has not been corroborated.If the correlation exists, free-market capitalist reforms may bear some of the responsibility for the phenomenon.

Materials and methods
I use a panel dataset covering 141 countries between 2006 and 2017, a significant period of the democratic recession, and fixed-effects ordinary least squares (FE OLS) regression analyses of democracy on Fraser's economic freedom index (F-EFI).As explained in more detail below, I also regress democracy on individual subcategories of the F-EFI.In the analyses, I control for a variety of potentially confounding variables, including country-fixed effects, to avoid omitted variable bias.The starting year was chosen as 2006 because it is the peak democratic year according to Freedom House data and Diamond's (2015) famed diagnosis of the democratic recession.The 2017 cutoff was necessary to ensure the most complete data coverage with respect to all the variables employed.My sample of countries has been drawn from the replication dataset in de Soysa and Vadlamannati (2021) and relies almost exclusively on countries with full data coverage over the examined period. 1 Freedom House (2022) democracy scores and World Bank (2022) data on population size have been imported into the dataset.

Independent variable
The main independent variable of interest, which has been lagged by 1 year, is F-EFI.Since 2000, the Fraser Institute publishes yearly data on economic freedom for most of the world's countries (Gwartney, et al. 2021).I have opted to use F-EFI instead of a competing index because it uses third-party data, minimizing the likelihood of the index being subjectively colored by the Fraser Institute's politics or ideology (which is a potential worry with the Heritage Foundation's index).The aggregate index, which is presented on a scale from 0 to 10 (10 being most economically free), is made up of the average score of 5 categories: size of government, legal system and property rights, sound money, freedom of international trade, and regulation.Each of these categories is further composed of individual indicators, as demonstrated in Table 1 below.
A higher (lower) aggregate economic freedom score therefore indicates that a country is closer to (farther away from) the free-market capitalist ideal.The same holds for the various measures of subcomponents of the index (the lower the taxes, tariffs, regulations, the higher the score).In 2019, the top 10 countries in the world by economic freedom were (in order) Hong Kong, Singapore, New Zealand, Switzerland, Georgia, United States, Ireland, Lithuania, Australia, and Denmark.All of the data are thirdparty sourced, not subjectively generated by the Fraser Institute itself.
It should be clear, both conceptually and empirically, that the economic freedom index manages to a large extent to capture what is typically meant with 'free-market capitalism'.It is regularly used in peer-reviewed journals for this purpose (for a sample of recent salient papers see the first section in the literature review above); the top 10 countries by economic freedom mostly align with the common intuition about which countries are more on the free-market side (say, Singapore, Hong Kong, the Anglo-Saxon world, and a few ex-communist, shock-therapy countries like Georgia and Lithuania); and the categories and subcategories mostly align with typical free-market capitalist characteristics like privatization, deregulation, and the shrinking of the state.Nevertheless, according to some critics (for instance, Burgis 2021), the aggregate economic freedom index is not a robust measure of free-market capitalism characteristics.Especially the legal system and property rights category is made up of several subcategories that are arguably not picking up distinct free-market institutions.Judicial independence, impartial courts, military interference in rule of law and politics, integrity of the legal system, and reliability of police all arguably do not have much to do with how free-market an economy is.The same could perhaps be said of the sound money category.To address this potential critique in my analysis, I disaggregate the index into its subcomponents below, and look at the correlation between each of them (not just the aggregate index) and democracy.

Dependent variable
The dependent variable consists of Freedom House (FH) aggregate data measuring the extent of political and civil rights in each individual country.FH reports yearly data on democracy for almost all countries in the world (Freedom House 2022).There are 10 indicators of political rights and 15 indicators of civil liberties a country is scored on.For each indicator, a country can be awarded 0 to 4 points (4 representing the greatest degree of freedom), for a maximum total of 100 points (representing the greatest degree of freedom).Political rights indicators are grouped into three subcategories: electoral process, political pluralism and participation, and function of government.Civil liberties indicators are grouped into four subcategories: freedom of expression and belief, associational and organizational rights, rule of law, and personal autonomy and individual rights.In 2019, the top 10 most democratic countries in the world were (in order) Finland, Norway, Sweden, Canada, Netherlands, Australia, Luxembourg, New Zealand, Uruguay, and Denmark (all scored above 97).

Control variables
A standard battery of controls is included in my analysis.Drawing on the de Soysa and Vadlamannati (2021b) dataset, I control for income per capita, growth rate of the economy, count of peace years, civil conflict (as a dummy variable), population size, and the ratio of oil rents to GDP. 2 These controls are typically used in studies of economic freedom, and they make theoretical sense for the present study.All of them are likely confounding factors which could be responsible for spurious positive or negative correlations between economic freedom and democracy.Income per capita (but also the growth rate, which captures the presence or absence of economic crisis) is expected to positively correlate with democracy, as predicted by modernization theory.Based on the literature review above it is also likely to be positively correlated with economic freedom.Conflict is likely to reduce chances of improvement of both economic freedom and democracy.Population size could be related to ease of economic and political governance, with smaller nations being easier to govern.A higher amount of oil rents indicates a potential resource curse, which makes democratization less likely.A log value of income per capita and population size has been used to reduce the role of extreme values, and both income per capita and growth rate variables have been lagged by 1 year.

Results
As is common with panel data containing a high N and short time-period, examination of OLS diagnostics suggests that both heteroskedasticity and autocorrelation (or serial correlation) are present in the data. 3To address both issues, I calculate and report robust standard errors which have been clustered within countries (Wooldridge 2019).These standard errors are robust both to heteroskedasticity and autocorrelation.There is also the potential issue of multicollinearity.However, the correlation matrix produces low variance inflation factor (VIF) scores, suggesting that multicollinearity is not a problem in the dataset.When using the aggregate F-EFI index, all variables' VIF scores are below 2, and when using the index's subcomponents all but one are below 4 (the legal system and property rights exception has a score of 4.126, so it is in the acceptable range of VIF <5) (Table 2).
Figure 2 shows the simplest regression of democracy on economic freedom in a single-year cross-section (2017) and without any controls.Mere visual inspection of the graph should suffice to demonstrate that the relationship between the two variables is positive and very strong (r ¼ 0.623), with 40% of the variance in democracy scores being accounted for by economic freedom scores.
Moving on to more interesting and complicated analyses, Table 3 reports fixed-effects panel regression results using the aggregate F-EFI index, as well as all of its individual 5 components, as the main independent variable of interest.All of the regressions in this table and Table 4 include time-fixed effects.Country-fixed effects are so far excluded; regressions incorporating country-fixed effects are reported separately below in Table 5.I have done so primarily for methodological reasons.Time-fixed effects enable us to peer into cross-sectional variation between countries (much like the simple regression in Figure 2), but they do so by averaging the cross-sectional variation using multiple years of data, not a single time point (unlike in the simple regression in Figure 2).In short, they tell us about cross-country variation averaged over many years.In contrast, country-fixed effects allow us to see the within-country variation in the variables of interest over time.This is an important methodological difference.One analysis tells us whether countries that are more economically free happen to also be more democratic (i.e., how different levels of economic freedom relate to different levels of democracy across countries), while the other analysis tells us whether countries that are increasing in their economic freedom over time are also growing more democratic (i.e., how changes in levels in the one variable relate to changes in the other within countries).Both questions are interesting and important, but for reasons of interpretation and understanding they should be treated separately.A recent paper on the limitations of fixed-effects models for panel data warns that 'it is hard to imagine many theories or empirical research questions that could be directly addressed by two-way fixed-effects coefficients.'(Hill et al. 2020:9) Similarly, Kropko and Kubinec (2020:12) warn that regressions employing both types of fixed-effects at the same time 'will often be difficult to communicate and to understand', because they consist of 'the average difference in intra-unit changes in the dependent variable at time point t, averaged across time points.'(Hill et al. 2020:9)    In the first regression in Table 3, only the F-EFI index and FH democracy scores have been used.In line with the simple single-year cross-sectional result from Figure 2, the regression shows that the two variables are statistically significantly related (p < 0.001) and that a one-unit increase in economic freedom (on 0-10 scale) is associated with an almost 20-unit increase in democracy (on a 0-100 scale), which is an extremely strong effect.
The second regression then employs all control variables, and it shows that the simple positive relationship between economic freedom and democracy is indeed confounded by GDP per capita, growth rate, civil conflict, count of peace years, and population size.However, economic freedom is still a statistically significant and substantively important predictor of democracy, although its coefficient size is notably smaller than in the first regression.A one-standard deviation increase in F-EFI results in about 1/8 of a standard deviation increase in democracy.GDP per capita is strongly positively associated with democracy, a finding in line with the classic literature on modernization theory butsurprisingly -not the contemporary literature on the qualified version of the theory, which finds a robust effect only with 10 to 20-year lags in the independent variable (see Treisman 2020).Also predictable, measures of present and past conflict have a negative relationship with democracy, as do oil rents.
The third and final regression in Table 3 uses individual components of F-EFI.It shows that only two out of five components, namely freedom of international trade and (lack of) regulation, are responsible for driving the relationship at the level of the aggregate index.Freedom of international trade has a positive and sizable effect on democracy (around a fifth of a standard deviation), while regulation has a negative effect (with the effect size of around 1/8 of a standard deviation).The three remaining components, legal system/property rights, size of government and sound money, are all statistically insignificant.These results indicate that while the overall association between economic freedom and democracy is positive, not all characteristics of economic freedom work positively.
Table 4 above displays the results of regressions employing the same control variables, but these regressions now also distinguish between developed and developing countries (using the standard criteria of classification by the IMF and World Bank).It is theoretically plausible that additional free-market capitalist reforms in already robustly capitalist economies have a different effect on democracy than the initial, more basic moves that developing countries are making from strongly controlled market economies to more normally liberalized economies.
The first two regressions from Table 4 use the aggregate F-EFI.As far as the main independent variable of interest is concerned, these regressions do not manage to completely replicate the positive result from before in both subsamples of countries.The positive association between economic freedom and democracy appears only in the developed countries.
The next two regressions from Table 4 employ data from the disaggregated F-EFI.It turns out that in the developed world, only two components exhibit statistical significance.The legal system/property rights component is positively associated with democracy and has a relatively large coefficient size, while size of government shows a negative association but has a miniscule coefficient.In the developing world, on the other hand, freedom of international trade and regulation are the only two statistically significant components, with the signs pointing in the positive and negative direction, respectively.Overall, the results indicate that the relationship between economic freedom and democracy is complex and multifaceted.I say more on this in the discussion section.
The robustness of these results has been verified by running regressions on an alternative measure of democracy.As can be seen from Tables A1 and A2 in the Appendix, the results remain almost completely the same.The effect sizes barely change, and all signs point in the same direction.One exception is the legal system/property rights component, which now turns significant both in the complete sample of countries and in the developing subsample, while before it was significant only in the developed subsample.The other exception is that size of government turns insignificant in the developed subsample.The broad picture of the relationship between economic freedom and democracy stays the same.
However, it is likely that there exist unobserved, time-invariant characteristics of individual countries which have an additional confounding effect on the relationship between economic freedom and democracy.That is why additional regressions with country-fixed effects have also been performed.As already said above, these regressions are not only useful because they are an additional remedy for omitted-variable bias (apart from the standard control variables) but also because they test the relationship between the two main variables of interest on the basis of within-country -not just between-country -comparison.This allows us to see whether changes in the level of economic freedom (not just the level itself) positively relate to democracy.Table 5 reports the results.
The first regression (Model 1) uses aggregate F-EFI and finds that the relationship between economic freedom and democracy is statistically insignificant.The second regression (Model 2) disaggregates F-EFI.Here, too, no individual F-EFI component stays statistically significant.
The next two regressions present the results when using the aggregate index in the developed and developing countries samples, respectively.In both samples, aggregate F-EFI turn out to be statistically insignificant.
The final two regressions again employ the two subsamples of countries but disaggregate the main independent variable of interest.There are only two statistically significant results, both in the developed countries subsample.Size of government exhibits a positive association with democracy, but the coefficient size is miniscule.Regulation (or lack thereof) exhibits a negative association but also with a small coefficient.
How well do these country-fixed findings stand up to an alternative measure of democracy?As can be seen in Table A3 in the Appendix, the three regressions using the aggregate F-EFI (total sample, developed subsample, developing subsample) remain insignificant, just as in Table 5.The only other two significant results are the negative coefficients of freedom of international trade and regulation in the developed countries subsample, but these are not large enough to show up at the level of the aggregate index.
Across all the different specifications then, the initial finding of a positive association between the aggregate F-EFI and democracy does not stay significant (although it never points in the negative direction).Specifically, the initial positive finding is sensitive to the inclusion of country-fixed effects.In other words, the finding shows up in the between-country analyses (based on time-fixed effects) but not within-country analyses (based on country-fixed effects).
Heterogeneity of results is also found when it comes to the disaggregated index across different regressions and samples.In the developing world, between-country analyses show that freedom of international trade is strongly positively associated with democracy (regardless of the measure of democracy), while regulation, or lack thereof, is moderately negatively associated with democracy.Between-country analyses in the developed world show that legal system and property rights are moderately strongly associated in a positive way with democracy.Importantly, however, none of these between-country findings stand up when performing within-country comparisons, which means that they are not robust to the inclusion of country-fixed effects.

Discussion
The debate on the social effects or correlates of free-market capitalist institutions has been ongoing for more than two countries, or even longer.On the one hand, critics following in the footsteps of Karl Marx's seminal analysis (for instance, Marx 1992Marx [1867]]), such as Harvey (2005), Klein (2007), Wright (2010), Bruff (2014), andBrown (2019), maintain that one of the pernicious effects of free-market capitalism has been the erosion of democratic rights, presently in the form of the democratic recession phenomenon.On the other hand, those with a more favorable or at least less critical disposition toward free markets, such as Norberg (2008) and Bjørnskov (2018) deny this and suggest that, to the contrary, there seems to exist a virtuous circle between capitalism and democracy, by way of which democracy is only possible in a capitalist society and at the same time capitalist growth is made more dynamic by democracy (Acemoglu et al. 2019) The results of the present study have, overall, been more mixed than would be expected on either side of the debate.My analysis indicates that free-market capitalism (as proxied by the aggregate F-EFI) does not have a negative, let alone strongly or obviously negative, relationship with democracy, at least in the period between 2006 and 2017, which represent a significant portion of the ongoing democratic recession.There is either a positive relationship (in cross-country analyses) or a neutral relationship (in within-country analyses) between the aggregate measure of free-market capitalism and democracy.This is in contrast to the hypothesis that free-market reforms have been strongly associated with the ongoing democratic recession.
Interestingly, however, the relationship between free-market capitalism and democracy turns out to be even more nuanced when the index is disaggregated.In the laxer specifications without country-fixed effects, free market institutions show different effects depending on the sample and the particular aspects of these institutions.First, secure property rights and a robust legal system are positively associated with democracy in the developed world but not in the developing world (depending on the measure of democracy).Second, freedom of international trade is positively associated with democracy in the developing world but not in developed countries.Third, modest regulation is negatively associated with democracy in the developing world, but seems to have a neutral relationship with democracy in the developed world.One would thus do well in agreeing with the conclusion that Graafland and Compen (2015: 807; see also McLean et al. 2019: 69) made in their otherwise unrelated study examining the relationship between economic freedom and life satisfaction, namely that 'the golden straightjacket that neoliberals propose … is [as far as democracy is concerned] not optimal in all respects.'However, because most of the enumerated components do not retain the same (or significant) sign in the more demanding specifications employing country-fixed effects, which significantly reduce omitted-variable bias, even such a conclusion might be too hasty.The paper has shown that, between 2006 and 2017, additional increases in economic freedom within countries have not been associated with additional increases (or decreases) of democracy.This is true both for the aggregate economic freedom index and its various subcomponents, and holds both for the total sample of countries and the developed/developing subsamples.The only exception is that additional deregulation reforms within countries of the developed world have been associated with (small) decreases in democracy in this time-period.
The overall mixed nature of the findings, especially when it comes to the disaggregated F-EFI, is not surprising either empirically or theoretically.
Empirically, the large literature review of the effects of economic freedom cited in the second section discovered that studies using the aggregate index are 'twice as likely to find a positive result as a mixed/null/uncertain result,' (Lawson, et al. 2022: 7) while studies employing subcomponents of the index 'were much less likely to find a positive result and much more likely to find a mixed/null/uncertain one.'Moreover, the review showed that of the more than 700 studies, almost 50% found a mixed/null/uncertain result.Recall also that in his study of how economic freedom impacts press freedom (an important component of democracy), Bjørnskov (2018) found that only 'market openness', but not size of government and regulation, has a significant and positive association with press freedom.
Theoretically, there are plausible mechanistic accounts of (parts of) economic freedom both aiding democracy and undermining it (discussed in the conceptual background section of the paper).This is completely compatible with the finding that certain aspects of economic freedom in certain groups of countries and regressions seem to have a negative effect on democracy, while aggregate economic freedom overall -that is, on net -relates either neutrally or positively, but not negatively, to democracy.

Conclusion
Generally, there are three general conclusions that can be drawn from the present study.
First, the idea that free-market capitalist reforms clearly and significantly undermine democracy in general, and have contributed to the ongoing democratic recession in particular, is not borne out in this study's sample and measures.In my regressions, aggregate economic freedom index is either positively or neutrally -but not negatively -related to democracy.
Second, developed countries with higher existing levels of free-market institutions have been and have remained more democratic throughout the period of the democratic recession under examination, mostly on account of their security of property rights and the robustness of their legal systems.The relationship is more complicated in the developing countries.Developing countries with higher existing levels of freedom of international trade, specifically, have been and remain more democratic than their more closed-off counterparts, while developing countries with higher existing levels of modest regulation have been less comparatively democratic.In this sense, free market capitalism in the developing world has an ambiguous relationship with democracy, although the relationship has been overall -on net -neutral so far.
Third, subsequent changes in levels of free-market institutions in both the developing and developed world over the 2006-2017 period have overall and in almost all specific aspects not been associated either positively or negatively with changes in levels of democracy.The overall relationship is neutral.Nevertheless, the evidence indicates that within the developed world, further deregulation in this period has been (weakly) associated with the erosion of democracy, which provides a limited amount of corroboration as far as the critical hypotheses about free markets and democracy is concerned.
Policy-wise, the results of the study suggest two implications.First, because free-market institutions cannot overall be related to the democratic recession in a notable way, the idea that rolling back neoliberal reforms would be a remedy for the populist onslaught on democracy seems dubious.It seems we do not have to pick between the robustness of democracy on the one hand and higher levels of free-market institutions on the other.Second, because there is some indication that, over the studied timeperiod, a reduction in regulation specifically was connected to a slight reduction in democracy, policymakers should consider this aspect of free-market capitalism as potentially problematic in light of the democratic recession.
There are a few noteworthy limitations of the study that could be addressed in future research.Due to data availability, the study was limited to only the first decade of the democratic recession (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017).Perhaps the connection between free-market capitalism and the democratic recession would turn out to be more complicated or even negative if the last several years of the democratic recession were included.Moreover, the study used two measures of democracy (namely, the one provided by Freedom House and Varieties of Democracy).More measures, say the Polity or the Economist Intelligence Unit indexes, could be used to further test the robustness of this study's findings.

Notes
1.I have recoded several countries marked as developing by de Soysa and Vadlamannati to reflect the fact that they were actually considered by both the International Monetary Fund and World Bank as developed for the whole (or the majority) of the 2006-2017 period.2. I have recoded the civil war dummy, which represented the US and Israel as experiencing a civil war between 2006 and 2017, to reflect the fact that they are not having a civil war.3. Heteroskedasticity means that the standard deviations of the dependent variable are nonconstant in relation to the independent variable, which -if not corrected by calculating heteroskedasticity-robust standard errors -is a violation of one of the basic assumptions of OLS.Autocorrelation means that the same variable is correlated with itself at multiple observations, which also has to be corrected for (in this case by clustering standard errors within countries).

Table 1 .
Disaggregated presentation of the independent variable (F-EFI).

Table 5 .
Unstandardized coefficients, 141 countries (2006Unstandardized coefficients, 141 countries ( -2017)), and developed and developing country subsamples according to the IMF and World Bank criteria; dependent variable: Freedom House democracy scores.

Table A3 .
Unstandardized coefficients, 141 countries (2006Unstandardized coefficients, 141 countries ( -2017)), and developed and developing country subsamples according to the IMF and World Bank criteria; dependent variable: V-Dem Liberal Democracy Index.