R Shiny Shows Decline in Even Strongest Democracies

Is Democracy in Trouble?

A simple question often asked is split into two parts: is my country a democracy? If so, is it a stable one? A democracy, if going by the Varieties of Democracy Project's definition of a liberal democracy, is defined as a country with a negative view of political power. It focuses on protecting individual rights against the state and the majority. This protection is achieved through a strong rule of law and checks and balances that both limit excess executive power. When one thinks of a strong, stable liberal democracy, countries that come to mind are normally those like the United States, Germany, Belgium, Denmark, France, and Japan. However, is it surprising to know that, since 2016, the democracy score for all of these countries have decreased? And this includes Denmark, which the V-Dem Project regards as the strongest liberal democracy in the world. What is this trend, though? (Featured img. credit Alisdare Hickson)

Democratic Backsliding: A Myth Becoming Reality

Democratic backsliding is, by most definitions, a decline in the institutions that make a country a democracy. For the V-Dem Project, it is an "episode of autocratization" where a country begins moving towards a more autocratic regime type. These regime types are "democracies", "electoral autocracies", and "closed autocracies". An autocracy has limited fundamental rights, such as freedom of expression and association, and multiparty elections either do not exist or are not free and fair (if electoral). Democracies can become autocracies gradually, year-by-year as they become more repressive, or it can be done instantaneously, such as through a coup. It is usually deliberate, either done by the elected officials or those trying to seek election.

The Peak of the Third Wave?

There were three "waves" of democracy". The first beginning with the rise of suffrage in 1828, the second after the end of World War II in 1945, and then finally with the fading of the Cold War with the collapse of autocratic regimes beginning in 1974. Each wave reaches a "peak", where the number of countries democratizing no longer increases. However, in our third wave the peak seems to be much more of a concern. Not only are countries no longer democratizing, but the number of countries autocratizing are outpacing that number. A belief that was previously held was that strong, wealthy democracies would remain so. Recent data trends, especially in the mid-2000s, began to show that this is no longer the case. In fact, third wave democracies, like Taiwan, have even begun to do better with democratization than their longer-lasting democratic counterparts. But what is causing this trend?

Right-wing populists Viktor Orbán (left) of Hungary’s Fidesz and Santiago Abascal (right) of Spain’s Vox. These two countries have recently experienced autocratization, with Hungary having changed to an electoral autocracy in the mid-2010s (Img. credit VOX España)

The V-Dem Project and Factors Influencing Backsliding

As mentioned, the V-Dem Project monitors the level of democracy of the countries in the world. It took note of democratic backsliding, how the rise in democracies since the beginning of the third wave is now being countered by a recent trend of autocratization. The V-Dem Project collects numerous categories on the elements of a democracy. In fact, "liberal democracy" is only one of the five indices used to define it, but it's one of the more comprehensive ones, which includes the "electoral democracy index".

Furthermore, the V-Dem Project looks at party identity and organization. For this, there are two indices. The first is "party populism", which is a rhetoric strategy of political parties that incorporate anti-elite, anti-corruption, and anti-establishment values. The second is "anti-pluralism", which is whether or not said political parties commit to democratic norms.

Some of the strongest factors influencing democratic backsliding determined by the V-Dem Project seem to be populism, political polarization, and inequality. A highly polarized society has opposite groups reluctant to engage with each other, perhaps even violently so. Inequality is typically measured through the Gini coefficient, which measures the income dispersion across a population.

President Recep Tayyip Erdogan (right) of Turkey presides over one of the most polarized countries in the world, and is often cited as one of the most egregious examples of backsliding in a country that was previously experiencing democratizing (Img. credit US Department of State)

Visualizing Democratic Backsliding: R Shiny

To better visualize what has been termed the "democratic collapse", an R Shiny project was created. Shiny is a framework made to build a web application, using R code. All key aspects of the V-Dem Projects findings were included. This includes the decline in democracy as a trend in itself, and then populism, anti-pluralism, and the Gini coefficient as factors.

The goal here was to A) confirm the results of general convention and B) create a reactive and interactive application that is easy to use. The entire project can be found on my git repository, along with the app link.

Backsliding as a Trend

Using Leaflet

Usually, the best way to visualize a trend on a national level would be to view some sort of map. The V-Dem dataset was mainly based on a timeline, although it used countries for an index. The problem was that this means that it didn't account for actual geographic location. Therefore, all countries being examined had to be turned into a GeoJSON file type, where each now would have latitude, longitude, and the polygons/its constraints. However, they no longer had the data that the V-Dem Project provided. So, it was inputted manually, based on the year constraints for how far back the Gini coefficients went. The tab itself allows one to hover over any country in the world and see its name and liberal democracy index, and then also switch to a different year as far back as 1964. However, notable recent downward trends began occurring mostly since 2008. A description subtab is also provided explaining the graph, where a score of 0 is a total autocracy, and a score of 1 is a perfect democracy.

Above shows the Leaflet map of the liberal democracy index. Canada, Hungary, and China are all used as an example of each different regime type.

A Downward Trend: Democracies

In addition to looking at the democracy index on a global level, a secondary type of graph was created where the countries were broken down into the previously mentioned regime types, and then further separated based on two qualifying factors: A) the country has been moving towards an autocracy since 2013 and/or B) the country was in a state of autocratization in 2022.

Shown are the most egregious examples of liberal democracies that experienced democratic backsliding. These are the countries that both moved towards autocracy since 2013 and were on a downward trend in 2022. The US recently recovered from backsliding, and is included to compare against.

A Downward Trend: Autocracies

The regime type categories are based on a V-Dem Project classification, and has no set value. There is a "gray area" between .3 and .5, where countries are sometimes considered electoral autocracies and "electoral democracies". This is based on variation between the indices that the V-Dem Project uses. For example, Poland has an electoral index of .59 and a liberal index of .44 in 2023, but is still classified as an "electoral democracy".

Pictured are the electoral autocracies, which have recently become more numerous and have had sharper declines than the liberal democracies. Selected once again are the most egregious examples, but all countries experiencing backsliding can be selected.

When it comes to gradual or sharp declines in democratic backsliding, one can most easily see gradual declines in electoral autocracies, and the sharp declines in the closed autocracies. For example, Hungary, classified as an electoral autocracy, was at .35 in 2021 and is now at .325 in 2023, and we know that it is a country both experiencing decline since 2013 and is in a state of decline. Burkina Faso, on the other hand, experienced a coup in 2022, going from .36 at the beginning of 2022 to .18 the next year.

The final graph of the trends shows the closed autocracies, which show the sharpest declines. Note that every country pictured, besides Iran, has either experienced an internal civil conflict or a coup recently.

Populism Favoring Autocracy

Party Identity as a Factor

Populism is directly correlated with anti-pluralism. As populist parties continue to use their rhetoric, they are more likely to be unwilling to engage with political opponents, and further demonize their opponents. Populism and anti-pluralism are their own indices, in a separate data frame, but this was combined to measure it against the liberal democracy index. The second tab shows two trends: the gradual rise in populism, and then a relatively steady rate in the increase and decrease of anti-pluralism.

Populist rhetoric has increased globally since 1970.

Anti-pluralism has maintained a relatively stagnant trend, suggesting it is always prevalent and its impact is stronger with other factors.

Significance of Populism

When analyzing party effect on the liberal democracy value, the parties were categorized by "government support", such as whether they were a senior partner/had ruling positions in a ruling coalition, or if they were in the opposition government. For populism, almost every grouping is statistically significant, with the exception of "in government, but not represented", which is a rarer case where the party is in the ruling coalition, but has no representation in parliament. The strongest relationship is if the party was in opposition, with a t-value of -13. However, the variances, which is how much these factors explain the relationship, for all groupings were very low (highest variance was .06). Interestingly, more senior partner populist countries are also more autocratic.

If the populist parties were in the government, or in opposition, then they all have a statistically significant negative relationship with the democracy value (Senior Partner (0), Junior Partner (1), In Opposition (3)).

The Rhetoric of Hate: Anti-Pluralism

When anti-pluralism is in taken into account, we see a much stronger relationship in all categories. The strongest relationship is now those parties that are in government (t-value of -62), and no relationship is statistically insignificant. Anti-pluralism also has much higher variance (highest being .70) than populism in all categories, with distributions better representing a linear relationship. One can select any amount of regression lines on both tabs, and select the line itself to see the strength of the relationship.

Anti-pluralism shows a stronger impact, with a statistically significant, negative relationship in call categories, including the previous non-significant category of "in government, but not represented (2)".

Tearing Ourselves Apart From The Inside

Political Polarization's Staggering Rise

For almost every country in the world, since the 2000s, political polarization has been increasing, and in some cases dramatically. This compares with populism, which has increased gradually, and anti-pluralism, which has relatively stayed the same. Polarization is concerning in that it represents society as a whole, reflected by the parties in the government and their views on democracy.

Selected are countries that have the highest levels of polarization in each political region of the world. Note how even a liberal democracy like the US has higher polarization than a more autocratic country like Zimbabwe, possibly explaining the backsliding in the country.

Polarization by Region: The Unique Threat

Polarization is not only statistically significant on a global level, but also on a regional level as well, in every region. Its strongest effects are shown on the European and Latin American regions, which tie at -30 in t-value, although Western Europe has a slightly strong variance at .40 compared to Latin America's .37.  Interestingly, regions that typically house more autocratic countries, like Asia and Eastern Europe, have notably weaker relationships and much lower variances, while the Arab Nations actually showed a reverse trend. This could be explained by how political expression being a requirement for polarization to matter, even if the country is an autocracy. This could be a cultural factor as well. Also, multiple parties need to exist for a society to be polarized, even if they are rubber-stamped like in China.

Western Europe (5) shows a negative linear trend of polarization, with extreme outliers possibly even further showcasing polarization's effects.

The Arab Nations (3) are unique in that, although still statistically significant, the trend is actually positive. It should be noted that many of these nations lack political camps to begin with, such as with Saudi Arabia, which is an absolute monarchy.

Explaining Backsliding Through Wealth Inequality

A Global Trend?

Usually, economic factors are the easier ones to point fingers at, and this one would be no different. It's been frequently stated that the rise in democratic backsliding in the developed countries became more prominent since the 2008 recession. Therefore, one would think that, as inequality in wealth distribution rises, the democracy level of the country would fall. In fact, looking at it on a global level, this does appear to be the case. And in countries that experienced sharp changes in their Gini coefficients, such as with Colombia, the change can be seen more drastically.

Shown is wealth inequality on a global level. The two factors reflect off of each other, where as Gini increases, the democracy score decreases, and vice versa.

The Weakest Factor

However, when the regressions were run by region, Gini coefficient consistently showed to have a weak relationship, although only Asia was statistically insignificant. Furthermore, the relationships themselves were erratic and had little sense of direction.  Only four of the six regions went in a very slightly negative direction, while the two others, Africa and the Arab Region, went in a positive direction. Even the strongest relationship, Western Europe, was notably effected by outliers when a weighted regression was applied, suggesting just how weak these relationships are. The explanation seems to be fairly simple: most countries do not have a Gini that changes enough for it to have a clear effect on the democracy index.

While Western Europe (5) has the strongest relationship at a -6 t-value, its variance was still very low at .06, along with it clearly being effected by outliers. Most countries in Western Europe have stable Gini coefficients, even though many can be high.

Africa(4) notably has a positive relationship instead, despite still being statistically significant. These countries typically have high levels of inequality, so it could mean that the factor is less relevant. For example, a highly democratic African country with poor wealth distribution would still be considered "successful".

Conclusion: The Threats to Democracy

From the data, it can be assessed that democratic backsliding has been occurring since the mid-2000s, even in countries that are considered established democracies. Going along with this backsliding, the level of populist rhetoric in political parties has been increasing gradually since 1970s. The political polarization of society has agreed with these trends as well, perhaps the most so as it has been shown to be consistently increasing globally. As probably expected, all of these factors have a negative relationship with the liberal democracy index.

Populism is statistically significant in all categories except "in government, but not represented". However, most of the relationships have a low variance, and the strength of the relationships are weak. Anti-pluralism, on the other hand, is statistically significant in all categories, have much stronger relationships, and its variance explains the relationships much more. Political polarization is also significant in all regions, but the stronger relationships are in those regions that already have more democracies. For example, we even see a reversal of the trend in the Arab countries.

Wealth inequality proved to be the lame-duck of the factors, despite previously being considered a strong factor causing democratic backsliding. On a global level there seems to be a noticeable relationship. However, when broken down by region, the strengths of the relationships are not only poor, but also lacking in consistency.

Protest conducted in 2018 by right-wing populist party Alternative for Germany. Germany is considered one of the strongest democracies in the world, and had considerably low polarization. However, like with almost all countries, polarization increased and its democracy score decreased, giving rise to extremism (Img. credit Matthias Berg).

Suggestions From A Technical Standpoint

Although highly interactive, some remarks can be made about the Shiny app itself. As noted, the Leaflet map was created by hand with manually inputted data. This means that the years were used as columns, instead of having "year" be used as a column, and just repeating the countries for each year. While simpler, it could be preferable to do it by row to allow for a slider for years, instead of having it be an input.

Further, instead of having the layout be based on tabs, a drop-down menu route could be chosen instead, along with having a theme. While each of the graphs themselves, including the map, allowed for some kind of input, alternative inputs can be used. For example, "clicknearby" can be used instead of "clickinput", or using "mousehover" like for the map for the graphs. Rather than having the descriptions be separate tabs entirely, adding more outputs could be suggested to add a "flow". This would be where the user could select certain parts of the data and be provided with information explaining just what they're looking at, rather than a summary of all the graphs in that tab.

Suggestions on Analysis

The goal of the project was essentially to see if the presumed "strongest factors" were correct. For the most part, they are, with a few caveats. This doesn't mean that the analysis itself cannot be improved. For example, the "liberal democracy index" was chosen as the main dependent variable, but there are five potential indices to choose from. All of these could even be combined, thereby using the mean instead. Furthermore, an arbitrary date was chosen based on the Gini coefficient data, where the V-Dem Project actually records data as far back as the 1700s. In fact, because of this, it could be suggested to use a different measurement for wealth inequality instead of the Gini coefficient considering that was the worst factor of the four.

Only the known most significant factors were chosen, and it might be odd to actually use an index against an index to begin with (in this case it was populism/anti-pluralism against democracy level). The V-Dem Project includes hundreds of factors, divided into multiple groups, each of which can be used as their own focus for a study. This includes elections and their qualities, the executive itself, corruption, and/or the legislature. This is useful, as the Shiny application was made so that it could be easily manipulated, where new data can replace the previous ones used.

Overall, these factors are too broad to make any real conclusions from in regards to a solution standpoint. Populism and anti-pluralism are indices in themselves, therefore effected by hundreds of other factors. For polarization, one would have to actually look at what's causing the factor to increase, since it's a societal issue. As for the Gini coefficient, it too is probably also more influenced by other factors when it comes to backsliding.