How Do Economic Conditions Influence Support for Democracy?

Abundant political economy work has found that societies tend to be less supportive of the political system that governs their country when the economy deteriorates. This paper seeks to enrich the existing literature on economic and democratic assessments by including the relatively under-explored group of developing countries in the Middle East and North Africa (MENA) region, many of which are Muslim countries. Using public opinion data from the World Values Survey of 78 countries from 2005 to 2020, the author applies multi-level regressions to test whether the relationship between democratic support and economic growth varies along three dimensions: country affluence, MENA versus non-MENA regions, and type of political regime. Consistent with past studies, the findings in this paper confirm that higher real gross domestic product (GDP) per capita growth significantly enhances public support for democracy. However, it also identifies that the more robustly significant predictors of public support for democracy are political variables, particularly evaluation of government performance. The interaction variables analysis reveals that the effects of economic growth on public support for democracy is weaker in the presence of higher country affluence as well as in more liberal democracies, while being located in the MENA region strengthens the relationship between economic growth and public support for democracy. In addition, respondents who consider free and fair elections as well as protection of civil liberties to be essential features of democracy tend to be more supportive of democracy itself. Meanwhile, those who associate democracy with redistributive policies tend to have a more negative assessment of democracy.


Introduction
Understanding sources of public support for democracy, including factors that can sustain or weaken public support for democracy amongst the electorate is vital for the durability of democracy itself.2A society that tolerates illiberal practices by the state carries a serious risk of democratic backsliding or outright reversion to autocracy.The issue is particularly important for newly democratized and democratizing regimes, which arguably cannot fully graduate to become mature democracies without citizens' steady support for democratic principles. 3or those who believe that a society judges its political regime by their leader's ability to deliver good performance, an important sub-set of the performance criterion is the economy. 4A poll by Pew Research Center conducted in 24 countries in 2018 documented that respondents who were dissatisfied with the economy were on average 36 percentage points more likely to be dissatisfied with democracy, compared to those with favorable views of the economy. 5In South Korea, the financial crisis of 1997-98 arrived amid a newly democratizing regime that had just toppled a military dictatorship, prompting many to blame democratization for the economic crisis. 6There is now abundant empirical evidence that worsening economic conditions are associated with lower support for democracy, although advanced economies are disproportionately represented in the sample. 7e know relatively less, however, about the precise magnitude of the economy's influence vis-à-vis political factors, such as the degree of political representation in influencing democratic support,8 and this question remains widely debated.One counterargument is that citizens actually value democracy for delivering political goods, such as representation and institutional processes that guarantee political rights and civil liberties, regardless of short-term fluctuations in economic performance. 9Moreover, existing studies tend to have limited geographical sample coverage; in particular, developing democracies and non-democratic countries are not well-represented, and neither are countries in Southeast Asia and the Middle East and North Africa (MENA).Expanding the sample coverage to include a larger number of newer democracies is important given the well-documented phenomenon of democratic backsliding among third-wave democracies, where public support for democracy is generally less stable compared to more mature democracies.
In order to specify when the economy matters more than political factors, it is important to recognize that a host of economic, social, and partisan issues work together in influencing public support for and satisfaction with democracy.This paper's main contribution lies in further exploring the conditional factors that amplify or weaken the relationship between economic performance and public support for democracy.
Using public opinion data from the World Values Survey, consisting of 78 countries and covering the period 2005 to 2020, the first part of the study addresses the baseline hypothesis: when tested over a wider range of countries (including authoritarian countries and nascent democracies in the MENA region), how sensitive is a country's support for democracy to the fluctuations of the national economy, as measured by real gross domestic product (GDP) per capita growth?In light of disagreements in the literature regarding whether economic factors are more dominant than political factors in explaining support for democracy or vice versa, widening the sample coverage becomes especially important.
Beyond enlarging the data set, this paper tests three country-level factors that may condition the effect of real GDP per capita growth on democratic support: 1) economy: country affluence or income level; 2) geography: MENA regions versus other parts of the world; and 3) political regimes: does the relationship vary between democratic and nondemocratic countries?Few would disagree that these are important factors to consider, yet existing work is limited and whether these factors amplify or weaken the baseline hypothesis remains the subject of much debate.Some academics argued that in newer democracies (many of which are developing countries), we should see a stronger connection between economic performance and support for democracy,10 while others argue the opposite, because of weaker economic accountability channels.11By including the above three conditional factors, this study aims to more fully appreciate the nuances in the relationships between public support for democracy among individuals in certain countries.
The main findings of the paper are that higher real GDP per capita growth significantly enhances public support for democracy, reaffirming past studies.However, when compared with political covariates, variables such as evaluation of government performance and the importance of living in a democracy are more robustly significant in predicting public support for democracy.Further analysis reveals that the effect of economic growth on public support for democracy is weaker in the presence of higher country affluence as well as more in liberal democracies, while being located in the MENA region strengthens the relationship between economic growth and public support for democracy.These findings suggest that the story is not all economics; politics also play an important role in studying sources of democratic support.
In the remaining parts of this paper, I discuss the existing literature on the effects of economic performance on public support for democracy.I then explain the hypotheses tested in this paper.The data and methodology are presented in the next section, followed by a discussion of the results of this study.The last section concludes, draws implications, and outlines future research agendas for this topic.
In order to specify when the economy matters more than political factors, we should recognize that citizens make distinctions between "baskets of political and economic goods" when assessing democratic institutions. 31As such, under certain conditions, the effect of one factor may dominate others.
In the second hypothesis of this paper, the author looks at an important yet undertheorized conditional factor: the level of country affluence or income.Existing work on this topic is still limited and largely argues that in developing countries we should see a stronger connection between economic performance and support for democracy, 32 given that economic issues tend to dominate public agendas during elections and key debates, to a greater degree than seen in many advanced economies.As such, it is argued that economic performance is a prime determinant of public support for democracy in developing and less-developed economies, but not in developed democracies.On the contrary, the average citizen in advanced economies tends to prioritize self-protection and selfexpression in evaluating democracy. 33Moreover, because voters in advanced economies in general have had a longer experience with democracy, they are thought to better understand that economic welfare may be an expected outcome of a democratic system, but not necessarily a feature of democracy itself.
However, the assumptions behind this reasoning are subject to further inspection.For one, it is not necessarily the case that the average voter in developing countries overemphasizes economic performance in assessing support for democracy.That is inconsistent with research that has found that, in developing countries, political rights and civil liberties are more important than economic considerations. 34Second, some studies within the economic voting literature suggest that accountability over incumbent performance is actually weaker in developing countries, as voters do not apply the rewardpunishment mechanism over economic performance as strongly as in more established democracies. 35By extension, this should mean that the correlation between economic 31 Christman 2018. 32Daoust and Nadeau 2021 found that citizens in newer democracies are more likely to dissatisfied about democracy when the economy weakens, compared to advanced economies and more mature democracies.Rohrschneider and Loveless 2010 argue that richer countries on average have voters who attach less importance and salience to materials concerns such as economic growth.Indeed, several public opinion surveys document that in developing countries, when asked what is "the most important problem" in your country, the economy appears front and center, whereas this premium on the economy is not as strong in advanced economies (e.g., Burlacu 2014). 33Inglehart and Welzel 2005. 34Bengtsson and Christensen 2016, Bratton and Mattes 2001; Braizat 2010. 35Bochsler and Hanni 2019.In fact, there is much debate about the role of economic considerations in political preferences in developing democracies.For one, the clientelistic nature of politics and remnants of conditions and democratic support should be weaker in developing countries.Third, just because in advanced economies multiple issues other than the economy occupy public airwaves and become publicly debated during elections, it does not necessarily mean that economic issues are not a main consideration for voters.Sometimes they are not, but other times they are, and quite strongly so.The results of the 2022 mid-term elections in the United States (US), for example, which took place amid elevated inflation rates not seen since the 1970s, seemed to suggest that the economy was a key factor on voters' minds. 36t is worth noting that many developing countries also happen to be younger democracies.In this fragile and early period of democratization, political elites often find ways to exert their power under the veil of democracy, with new regimes that take power after transitions becoming marred by rampant corruption.This was observed in Central and Eastern Europe after the fall of communism in the early 1990s 37 and in the Arab Uprising countries in the 2010s, 38 contributing to citizen skepticism of democracy, as democracy becomes seen to be associated with corruption and bad governance.
In sum, the second hypothesis tested is whether the relationship between economic performance and public support for democracy differs between high-income versus other countries that are considered developing countries.

MENA vs non-MENA Regions
The third hypothesis tested in this paper is whether countries in the MENA region display significantly different behavior than other countries in the sample.Specifically, I test whether the relationship between economic growth and support for democracy is stronger in MENA countries.
With a few exceptions, research on the effects of economic conditions on support for democracy has not included a notable amount of MENA countries in the sample, although there remains a strong interest in investigating how citizens in these Islamic countries view and understand democracy, especially since the Arab uprisings that called for democratic change just over a decade ago.Despite wide-spread protests in early 2010s, today only Tunisia is classified an 'electoral democracy', according to the data from the Varieties of Democracy project. 39Scholars have shown that support for democracy from 2011 to 2013 indeed declined in both Tunisia and Egypt and that these declines are largely explained by dissatisfaction with the government's political performance (e.g., corruption) and citizens' evaluation of the national economy. 40Additionally, results from a 2022 Arab Barometer survey 41 document that in most of the surveyed countries, more than half of authoritarian regimes (citizens with low political knowledge, low trust in democratic institutions, and inchoate party systems) weakens the accountability channel, casting doubt on the strength of economic voting.See Amri and Bouvet 2022 for a review. 36Horsley 2022. 37Roberts 2008. 38Kilavuz and Sumaktoyo 2020. 39Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman, Bernhard et al. 2021. 40 respondents agree with the statement that the economy is weak under a democratic system.More than half of the respondents in each country claim that they are more concerned about the effectiveness of their government's policies than they are about regime type.
Clearly, there are important dynamics in public opinion on democracy in this region and thus including MENA countries in the sample will enrich our dataset.Since citizens of MENA countries seem to care more about effective governance, it would appear that the average citizen has an instrumental understanding of political regimes and would presumably assess democracy based on the purported economic benefits that such a system would bring.Therefore, we should expect to see a stronger correlation between economic performance and support for democracy in MENA countries compared to more advanced democracies.However, it should be recognized that bringing in the MENA region into the analysis means there is more heterogeneity to the data.
Another idiosyncratic element of the MENA countries is that demand for democracy in nations such as Tunisia and Egypt in early 2010s took place amid a weak economic performance.This is the opposite of the supposed common path to democratization in Western democracies, where it is economic prosperity that gives rise to demands for democracy.Moreover, often the regime transition observed in MENA region is imposed by external actors (such as international organizations) rather than being the result of a home-grown movement. 42With little to no ownership of democracy at the roots, it would be logical to ask whether citizens in these countries tend to over-blame democracy when the national economy falls.

Political Regimes as A Moderator
The fourth hypothesis tested involves the type of political regime itself: whether the connection between economic performance and support for democracy differs between democratic and autocratic regimes.
Most existing studies estimating public opinion on democracy only include democratic countries.To an extent this makes sense, since the theoretical basis of this relationship between economic conditions and democratic support is the performance accountability criterion.When citizens are satisfied with the country's economic situation, they are more likely to support the incumbent and credit economic success to government policies, and by extension, they are more likely to think favorably about democracy itself.It is implied that such expressions of accountability are only viably present in a democracy. 43However, excluding authoritarian regimes in our analysis misses several important points.
First, many autocratic countries have felt pressure to democratize (such as during the Arab Spring revolutions in the 2010s) with varying outcomes, while unsettled or illiberal democracies often find themselves slipping back into authoritarianism mode (for example, Russia, Turkey, and Hungary).Therefore, not including non-democratic countries in the analysis misses the opportunity to better understand these dynamics. 42Burnell 2013. 43Dahlberg, Linde, and Holmberg 2014; Krieckhaus, Son, Bellinger, Wells.2014.Second, there is important research that shows that the average citizen in authoritarian regimes understands democracy differently from those living in democratic regimes. 44For instance, in non-democratic countries, economic or material considerations may be more important than procedural fairness or political processes when assessing democracy.If anything, the phenomenal economic rise of China and Vietnam, two autocratic regimes, over the past few decades have resulted in questions around whether democracy can be bypassed to achieve higher incomes.This paper therefore asks whether the economy is more likely to be a more dominant explanatory factor for democratic support if we include both democratic non-democratic countries in the sample, but differentiate the analysis.
Since most studies of public opinion on democracy only look at a sample of democratic regimes, the empirical analysis has differentiated between nascent versus established democracies, or between countries with different levels of democratic quality.One study 45 found that objective economic performance matters more than political performance in explaining public support for democracy; the same study claims that there is no evidence that citizens in non-democratic regimes will be more supportive of democracy when the regime performs well in economic terms.Focusing on satisfaction with (instead of support for) democracy, another study46 found that while economic performance is the best explanation for public satisfaction with democracy, the level and quality of democracy turn out to be a stronger predictor of differences in satisfaction with democracies that are observed between countries.
This study differentiates political regimes by using a dummy variable that equals to 1 for countries that are classified as liberal democracies and to 0 if the political regime is a closed autocracy, electoral autocracy, or electoral democracy, following the classification from the Varieties of Democracy project.Why is it important to compare liberal democracies to other types of political regimes?Although past research on this is still limited, the above literature survey suggests that there is little distinction between autocracies and newer democracies when it comes to expectations of what the political system should deliver, which is a growing economy.In other words, the maturity of democracy matters in how citizens evaluate and assess political institutions.While mature democracies also experience the ups and downs of the business cycle, unless there are systemic financial crises, generally citizens remain supportive of democracy.This may be because citizens feel that democracy gives voters a way to hold incumbent governments responsible for economic downturn, thus separating between demanding accountability from politicians and blaming the political system outright for poor economic performance.
An alternate way to put this is that voters in liberal democracies have a better understanding that while it is important to demand accountability for their leaders, voters should be able to separate between support for the incumbent government, which is affected by short-term performance such as the economy, and support for democratic principles, which is based on a more diffuse process, and not be over-reliant on the performance of the economy. 47

Data Description and Methodology
To test the above hypotheses regarding the relationship between national economic conditions and public support for democracy among individuals in a certain country and time, I use public opinion data from the World Values Survey (WVS) 48 database.I use the 5 th , 6th and 7 th waves of the WVS, covering years 2005 to 2020.The country-level data are sourced primarily from the World Bank and the Varieties of Democracy project.
I use a multi-level regression model to test whether the probability of being supportive of democracy is significantly related to the national economic performance, as measured by the growth of GDP per capita, controlling for individual-level political factors and socio-demographic variables, as explained in more detail in Table 1.For the data structure used in this paper, where multiple individual respondents are embedded or nested within countries, scholars have argued that multilevel regressions produce more reliable and unbiased estimates of the coefficients, compared to ordinary least squares (OLS) models. 49Multi-level regression models distinguishes between individual-level analysis and country-level analysis and recognizes that the relationship between the two main variables of interest (in this case: public support for democracy and national economic conditions) could depend on country-level characteristics, such as the type of political regime.More specifically, the multi-level model I employ is the random intercept model. 50hat is, the intercepts are allowed to vary by country, to account for the possibility that individuals in some countries may have systematically higher (or lower) levels of democratic support than in other countries.
A full list and description of the variables used in the analysis is available in Table 1, while summary statistics are included in Table 2. 47 Kotzian 2011. 48WVS is a research institute that provides data on socio-cultural and political change in over 90 countries from 1981 to 2020.The surveys are designed to obtain information about factors that influence intergenerational changes such as political factors (e.g., views on democracy), culture (e.g., trust), economic life, religion, gender roles, family norms, science and technology, etc. Local research institutes carry out in-person and/or phone surveys.For each country, a minimum of 1,000 respondents are sampled a stratified random sampling drawn to be representative the population along key socio-demographic strata.To measure public opinion towards democracy, I use a question item from the WVS module which asks respondents whether they think of democracy as a fairly good/very good or fairly bad/very bad way of governing their country.Originally an ordinal variable from 1-4, I converted this into a binary (0-1) variable for ease of interpretation.This item is commonly found in the democratic support literature and has been interpreted as a measure of the public's support for democracy, although it does not measure how satisfied respondents are with the performance of democracy in their respective countries. 52Due to the binary nature of the dependent variable, a multi-level logit model is applied.As can be seen from Table 2, support for democracy is generally quite high in this sample.The average score is 88%.
To capture the economic situation in the respondents' respective countries, the main independent variable, I use the growth rate of real GDP per capita, taking the values that are contemporaneous to the survey year in each country.Using an objective economic indicator such as this lowers the potential problem of reverse causality that can be with subjective indicators of the national economy.In that case, an individual who supports democracy may have an inherent bias of viewing the national economy more favorably.
The control variables are at both the country-level and individual variables.At the country-level, I control for the political regime type (liberal democracy versus autocracy, electoral autocracy, and electoral democracy) and a measure of executive corruption, with the idea that less corrupt politicians should, with all else equal, lead to more support for democracy.
In addition to individual political control variables (e.g., importance of living in a democracy, and how interested in politics are the respondents), the regressions also contain socio-demographic variables potentially linked to assessments of democratic preference.These include age, education level, and gender.
I use a set of control variables that explore respondents' preferences regarding certain features of democracy deemed most essential.This is a unique feature of WVS for waves 5, 6 and 7 that should be further explored. 53While democracy has many different features, research has found that different people tend to prioritize different features of democracy. 54Given the availability of the questions in WVS, I compare three different groups and test whether support for democracy varies among those groups: 1) those who primarily associate democracy with free and fair elections, 2) those who associate democracy largely with guarantees of civil liberties, and 3) those who have a view that in a democracy, the government shall redistribute income from the rich to the poor. 52 In total, the sample consists of about 176,785 individuals (in the baseline model, column 5 of Table 3) in about 78 countries, covering the period 2005-2020.The geographical spread included East-Asia and Pacific (10 countries), Europe and Central Asia (25 countries, including eight Western European and Nordic countries), Latin America and the Caribbean (10 countries), and Middle East and North Africa (10 countries).In addition, there are two North American countries, two South Asian countries, and eight countries from Sub-Saharan Africa.A full list of the countries in this study is available in the appendix.
To test the first or baseline hypothesis of the paper, I regress the binary dependent variable, support for democracy on economic growth (proxied by annual percentage change in real GDP per capita), controlling for political and socio-demographic control variables, including regime type and country-level corruption indicator.Methodologically, hypotheses 2, 3 and 4 (the conditional hypotheses) require interaction terms to capture differences in the effect of economic growth on the dependent variable across different sub-groups.Therefore, I use an interaction term between economic growth and: 1) the level of country affluence (as measured by real GDP per capita), 55 2) a dummy variable that equals to 1 if the country is in the MENA region, and 3) a dummy variable that equals 1 if the country is considered a liberal democracy. 55Following Daoust and Nadeau 2021.

Baseline Results
To test hypothesis 1, in Table 3, I first estimate the regression model with only the main independent variable, the growth rate of real GDP per capita (model 1 or the 'null model'), before adding control variables regarding different aspects of democracy deemed most essential56 (model 2).I then add 'political controls' (the importance of politics, the importance democracy, and the performance of the government) to the specifications (model 3) and finally, model 4 gives the full range of country-and individual-level covariates discussed in the previous section.Model 4 adds socio-demographic controls to the regression model.
The main takeaway from the baseline empirical results is that there is support for hypothesis 1.In the null model, which is similar to a bivariate regression, the coefficient on real GDP per capita growth is positive and significant at the 1% level.However, as more economic and political control variables are added, the statistical significance of the economic growth coefficient declines and it is not significant in model 2. In the model with the full set of control variables (model 4), real GDP per capita growth is positive and statistically significant at the 5% level.In sum, these results show that when the sample coverage is widened, as this study attempts to address, the positive effect of economic growth on democratic support still seems to hold, consistent with past studies.
The results confirm that a wide range of political and demographic variables also have a significant effect on democratic support in the expected direction.Demographically, older and more educated citizens are more likely to support democracy, consistent with past studies. 57s for political control variables, not only are the coefficients robustly significant across all models, but the estimated coefficients are also larger than the GDP per capita growth.Individuals who rate government performance favorably, who consider it important to live in a democracy, and who have higher levels of political interest are more likely to be supportive of democracy.Interestingly, perception of household financial situation is not significant, although considering the economy to be the most important problem is a significant predictor of support for democracy.As for the country-level covariates, the liberal democracy dummy variable is positive and highly significant, consistent with other studies, suggesting that democratic maturity is positively associated with democratic support. 58Political corruption has a negative and significant effect on support for democracy, but only in one specification (model 2).
Other interesting findings concern variables that capture respondents' views on the features of democracy considered most essential.Table 3 shows that, on average, the more highly a person rates free and fair elections and civil rights as important aspects of democracy, the more supportive they are of democracy.Conversely, those who associate democracy with redistributive policies tend to be less supportive of democracy, although this finding is not robustly significant.These findings imply that citizens do have different expectations of democracy and these differences have important political consequences, including for how they evaluate democracy itself.While these are interesting in their own right, I leave a more in-depth investigation into this question for future research.
In sum, consistent with past studies, this paper finds that higher economic performance significantly enhances public support for democracy.However, the more robustly significant predictors of public support for democracy are actually political variables, particularly evaluation of government performance, the importance of living in a democracy, and citizens' level of political interest.

Results for conditional hypotheses: country affluence, MENA region, and democratic regimes
In table 4, I ran multi-level logit regressions to test hypothesis 2 on countryaffluence (model 1 and 2) and hypothesis 3 on the comparison between MENA and non-MENA regions (model 3 and 4), while hypothesis 4 compares liberal democracies with autocracies (closed and electoral) and with electoral democracies (model 5 and 6).Models 1,3, and 5 include all country-and individual-level control variables minus demographic factors, while models 2,4, and 6 contain the full battery of control variables.All models include an interaction term between the main independent variable, real GDP per capita growth or economic growth, and the hypothesized conditional factor.
With regards to the second hypothesis, I test whether the relationship between economic growth and support for democracy varies based on country affluence, which is measured in terms of the level of real GDP per capita. 59This is captured with an interaction term between the growth of real GDP per capita and the level of GDP per capita.The existence of a significant interaction term would tell us, for example, that a 1% growth of real GDP per capita has a different meaning for support for democracy in a country where real GDP per capita is $50,000, compared to a country where real GDP per capita is only $5,000.The results in Table 4 suggest a negative interaction term, which reveals that the effects of economic growth on public support for democracy are weaker for higher income countries.However, this only holds for model 1, where demographic control variables are not included.Therefore, although not robustly significant, the findings suggest that in higher income countries, citizens' support for democracy is less sensitive to the short-term fluctuations of the national economy.It appears that voters in higher income countries do not over-blame the political regime when the economy is not performing well, which is consistent with the findings of past studies.
To test hypothesis 3, I use an interaction term between the MENA dummy variable and real GDP per capita growth.Models 3 and 4 suggest that this interaction term is positive and significant at the 1% level.This suggests that in MENA countries, the effect of economic growth on support for democracy is stronger compared to in non-MENA countries, which is consistent with the expectations outlined in the literature review.However, the coefficient on real GDP per capita growth itself is no longer significant, which means that the results from the interaction term alone is less robust.Interestingly, the MENA dummy coefficient is positive and significant, suggesting that there is a higher degree of public support for democracy.
Lastly, to test hypothesis 4 on whether the effects of economic growth on public support for democracy is conditioned by the type of political regime, in model 5 and model 6, the regressions include an interaction term between real GDP per capita growth and a dummy variable to capture liberal democracies.As discussed previously, the regime type dummy was chosen given that theory suggests that in autocracies and illiberal democracies, voters tend to associate democracy with economic welfare and conflate 59 Daoust and Nadeau 2021 also use the level of real GDP per capita to measure country affluence.For robustness, I replace the level of GDP/capita with a high-income dummy, and the results also suggest a positive coefficient on real GDP growth per capita.However, the interaction term is not significant.
dissatisfaction with a government's economic performance as being the same as dissatisfaction with democracy itself.The liberal democracy dummy was consistently positive and significant at the 1% level for all specifications, suggesting higher support for democracy in liberal democracies compared to other regime types.In addition, the interaction term between liberal democracy and real GDP per capita growth is negative and significant at both 5% (model 5) and 10% (model 6).This suggests that the effects of economic growth on public support for democracy are weaker in liberal democracies compared to the alternative political regimes considered here.Meanwhile, in models 5 and 6, the coefficient on real GDP per capita growth remains positive and significant at the 5% level.The results suggest that in liberal democracies, voters realize that they should not significantly alter their support for democracy every time the economic performance changes.Meanwhile, in other types of political regimes, there is a stronger expectation that democratic regimes should deliver economic welfare.
It is worth mentioning that the results from Table 4 regarding the sign and statistical significance of the control variables are consistent with the results of the baseline results in Table 3, which is that a wide range of political and demographic variables also have a significant effect on democratic support in the expected direction.Particularly notable is that the political variables -ratings of government performance, the importance of living in a democracy and political interest -are consistently significant at the 1% level in Table 4, and the magnitudes seem to outweigh the economic growth coefficient, just as in Table 3.Additionally, certain essential features of democracy correlate significantly and positively on support for democracy, such as the electoral democracy and liberal democracy features.Meanwhile, respondents who think that, in a democracy, the state should tax the rich and subsidize the poor, tend to be significantly less supportive of democracy than those who think otherwise.

Conclusion and Future Research
This paper contributes to the debate on the relative influence of economic factors in explaining public support for democracy vis-à-vis political variables.Specifically, the main contributions are twofold.One, when tested over a wider range of countries (including authoritarian countries and nascent democracies), the paper reaffirms that objective economic performance on average has a positive effect on support for democracy across countries and time, and also across individuals.
Two, this study uncovers that the story of sources of democratic support is not only an economic story, but also a political one.When looking at robustness of the results across different specifications, the variables that are most consistently significant are political variables, especially individual assessment of government performance.The importance of politics is further supported by the findings that the effect of real GDP per capita growth on democratic support is significantly weaker in higher-income countries (versus lower-income countries) and in liberal democratic countries (versus autocracies and electoral democracies).The results imply that in autocracies and illiberal democracies, the economy is more influential in explaining support for democracy.One can speculate that these findings suggest differences in understandings of democracy among individuals, a scholarship that has recently gained traction. 60In the future, it would be fruitful to test whether citizens in lower-income countries and less mature democracies have a tendency to focus on economic welfare as an essential feature of democracy, which would explain the tighter connection between economic performance and democratic support in this group of countries.Although economic welfare is a reasonable expected outcome of a democratic system, it is not a feature of democracy itself, and therefore voters should not over punish democracy every time economic performance changes.
This study undoubtedly has limitations, but it also opens up some new questions.It should be acknowledged one cannot fully analyze the dynamics of changing democratic support in the MENA region simply by including a country-dummy for location.Future work may need to account more specifically for which political or economic dimensions in the MENA region need to be considered in explaining changes in public support for democracy.These include not just the state of the political regime, but also the quality of democracy and the length of the country's experience with democracy (i.e., democratic maturity).Similarly, to capture the country's political regime only by a binary variable is an oversimplification, because democracy is a complex concept with a wide variety of meanings.
Related to the above, future research could include estimations using a different type of multi-level model.A random coefficient model allows the slope of the relationship between GDP per capita and public support for democracy to vary across countries.That is, it allows for the fact that in some countries, the relationship between GDP per capita growth and democratic support may be a positive one, while in others the opposite would hold.Similarly, more explanation is needed to better understand why certain features of democracy deemed essential by the respondents are positively or negatively correlated with support for democracy.This research has implications on understanding sources of democratic support among citizenry.If public support for democracy is overly sensitive to the ups and downs of economies, this is not ideal, as a political system should not be solely judged based on its economic output.After all, there are many external factors that influence economic outcomes that are outside of the agency of the government.Therefore, it is within the interest of political elites to ensure that citizens have a comprehensive understanding of democracy and do not over-emphasize the economic dimensions of democracy.