Pandemic Backsliding: Democracy During COVID-19 (March to September 2020)

The Pandemic Backsliding Project tracks state responses to Covid-19 and their potential effect on the overall quality of democracy within the country. The current version of the data reflects the situation between March and September 2020.

The Pandemic Democratic Violations Index (PanDem) captures the extent to which state responses to Covid-19 violate democratic standards for emergency responses. The Pandemic Backsliding Index (PanBack) reflects the extent to which such responses pose a risk to the overall quality of democracy within the country. Combined, these two indices provide a snapshot of how emergency responses to Covid-19 may be affecting the quality of democracy within the country. They are not intended to estimate the level of democracy, which is instead captured using the V-Dem Liberal Democracy Index (LDI) in 2019.

Use the options below to compare the country-level scores for the Pandemic Democratic Violations Index (PanDem), the Pandemic Backsliding Index (PanBack), and the Liberal Democracy Index (LDI) in 144 countries.


The map displays the scores for the selected index. Numbers in parentheses indicate the change in score from Q2 (Mar/Jun) to Q3 (Jul/Sep).
Additional country-level information is available below by clicking on the map.


Select a country from the map above for case-specific information.


Notes:
The 'March-September' time period reflects the worst (maximum) violation observed across the time period. For the PanBack index, we use the mean of the coded time periods to indicate the average risk of backsliding during the pandemic.
For the full text of comments and the list of sources, please see our Github page .

The Pandemic Backsliding Project tracks state responses to the Covid-19 pandemic as illustrative of the varieties of emergency measures and their execution, addressing how these decisions affect short- and long-term prospects for the political regime and democracy.

This Policy Brief summarizes key insights from our analysis.

Team and suggested citation

The Pandemic Backsliding Project is led by a team of four Principal Investigators - Anna Lührmann, Amanda B. Edgell, Seraphine F. Maerz, and Jean Lachapelle - and the Research Coordinator Sandra Grahn. We are based at the Varieties of Democracy Institute (V-Dem) at the University of Gothenburg in Sweden.

Suggested citation:
Edgell, Amanda B., Anna Lührmann, Seraphine F. Maerz, Jean Lachapelle, Sandra Grahn, Ana Flavia Good God, Martin Lundstedt, Natalia Natsika, Palina Kolvani, Paul Bederke, Shreeya Pillai, Abdalhadi Alijla, Tiago Fernandes, Hans Tung, Matthew Wilson and Staffan I. Lindberg. 2020. Pandemic Backsliding: Democracy During Covid-19 (PanDem), Version 4. Varieties of Democracy (V-Dem) Institute, www.v-dem.net/en/our-work/research-projects/pandemic-backsliding/

Methods

The Pandemic Backsliding Project bases its coding primarily upon data collection by a team of trained research assistants. The sources are documented at GitHub and include mainly official government sources, academic databases, trusted inter-governmental, state or independent organizations and trusted media outlets. In general one coder was assigned to one country, but for some observations two coders provided input and the principal investigators reconciled the information in cases of disagreement. Country experts, regional experts or the principal investigators have reviewed the coding that informs the Pandem-Index for most countries.

The full documentation of coding, data sources, and comments for each country and variable are accessible online at GitHub . Questions were designed in a way that facilitated the collection of factual information. Nevertheless, they involved some judgement on the side of coders. If in doubt, coders were instructed to consult with others and code something as less (rather than more) severe. This means that if our data errs, it errs on the side of under- rather than overreporting. Violations might have occurred that we do not capture, but we are quite confident that the violations we report have occurred.

Index construction

Pandemic Democratic Violations Index (PanDem)

Question: To what extent have government responses to the Covid-19 pandemic violated democratic standards for emergency measures?

Clarification: With this index, we aim to capture the degree to which democratic standards for emergency measures have been violated between 11 March 2020 and late September 2020 by government responses to the Covid-19 pandemic. Specifically, we draw on international agreements about the democratic standards for emergency measures (e.g. ICCPR) to categorize violations (see Table 1 below). Emergency responses are then coded as no violation (0), minor violation (1), some violation (2), and severe violation (3) in each of these sub types.

Scale: Interval, from low to high (0-1).

Aggregation: This index is formed by taking the sum of all violations recorded in Table 1. We then rescale into a 0-1 interval by dividing by the total possible additive score.

Pandemic Backsliding Index (PanBack)

Question: To what extent has democracy receded due to government violations of democratic standards in response to Covid-19?

Clarification: This index estimates the amount of democratic backsliding during the covid crisis. It builds on existing scholarship showing that a country's vulnerability to democratic backsliding is non-monotonic in the level of democracy. At low levels of democracy, there is little room for backsliding; and at high levels, the institutions of democracy are robust. The most vulnerable countries for backsliding are those in the mid-range, namely those that are neither fully democratic nor autocratic. The PanBack index captures such inverted U-shaped vulnerability to backsliding by weighing more heavily violations recorded in countries in the mid-range of V-Dem's Liberal Democracy index v2x_libdem measured in 2019.

Scale: Interval, from low to high (0-1).

Aggregation: The PanBack index is calculated as follows: PanBack = 4*PanDem_alt*v2x_libdem(1-v2x_libdem) where the factor of 4 is included to ensure the index ranges between zero and one.


For more details see our codebook.

Acknowledgements

We are grateful for the support of our research assistants -- Ana Flavia Good God, Martin Lundstedt, Natalia Natsika, Palina Kolvani, Paul Bederke, and Shreeya Pillai. We also thank Abdalhadi Alijla, Tiago Fernandes, Hans Tung, Matthew Wilson and Nina Ilchenko as well as V-Dem country managers Ane Mary Tusingwire, Chung Nguyen, Cláudia Araújo, Dina Milovanovic, Iqbal Ahnaf, Janiel Hazle, Jessie Moritz, João Cancela, Johanna Peltoniemi, Jonas Lefevere, José Santana Pereira, Kalilou Sidibe, Kevin Mazur, Kharis Templeman, Kimlong Chheng, Mahamane Yahaya, Mantobaye Moundigbaye, Marijn Nagtzaam, Nwana Collins, Ozan Utku Cam, Sahadevan P., Stephen Kini, Paolo Sosa, Randrianja Solofo, Ravi Dutta, Sitara Noor, Tahir Kilavuz, Yi-ting Wang, and Yoshikuni Ono for their input on V3 of the dataset. In April we conducted a pilot study and would like to thank Abdalhadi Alijla, Vanessa Alexandra Boese, Tiago Fernandes, Adea Gafuri, Dominik Hirndorf, Christopher Howell, Yuko Kasuya, Juraj Medzihorsky, Asma Shakir Khawaja, Carlos Shenga, Medet Tiulegenov for their input to the pilot version of this data set. We also would like to thank Laura Maxwell for the creation of this dashboard.