Monday, March 30, 2020

Air pollution in Tbilisi nearly halved by Covid-19 measures

[Note: This article was co-published with OC Media. It was written by Ian Goodrich, a policy analyst at CRRC Georgia. The views presented in this article do not necessarily represent the views of CRRC Georgia or any related entity.]

Particulate matter in Tbilisi’s air has fallen by as much as 45% following the introduction of measures to combat the spread of COVID-19, according to analysis of air quality data by CRRC Georgia.

The findings reflect broader global trends which have seen dramatic decreases in air pollution levels in China, Italy, and the United Kingdom.

Data from the Ministry of Environmental Protection and Agriculture show a clear fall in air pollution in the Georgian capital.

The plot below examines overall pollution levels in Tbilisi over the last month, overlaid on the same period for the last three years.

It highlights key dates in the COVID-19 crisis, specifically the first registered case, on 26 February, the closure of bars and restaurants and restrictions on entry into the country on 16 March, and the declaration of a state of emergency on 21 March.



Note: Two-day rolling average of the mean of normalized values for carbon monoxide, nitrogen dioxide, sulfur dioxide, and particulate matter (PM2.5, PM10).

Total pollution appears to be lower following the closure of bars, restaurants, and borders. The data also suggest that in the case of these closures, falls in pollution appear to have preempted policy decisions. The pattern that emerges should be intuitive for anyone who has looked outside in Tbilisi over the last few weeks.

Air pollution is however strongly seasonal, peaking and falling throughout the day, week, and year. It is also closely tied to weather patterns: strong wind, for example, will disperse pollutants. Modelling allows these factors to be taken into account when determining the overall impact of Covid-19 measures.

Models have been created for levels of five key pollutants in Tbilisi, examining particulate matter (PM 2.5 and PM 10), carbon monoxide, nitrogen dioxide, and sulfur dioxide. The models adjust for seasonal factors using Facebook’s Prophet tool, and weather using daily data from NASA.

The models show that following the declaration of emergency, almost all categories of air pollution fell. The only exception to this pattern is sulphur dioxide which has remained relatively constant throughout.



The most dramatic impacts from COVID-19 related measures are seen for particulate matter pollutants (PM2.5 and PM10) with each falling by 40%–45% after the emergency declaration, approaching half their pre-crisis rate.

Interestingly, for these substances, a sharp fall was present prior to the introduction of emergency measures. It is possible that this drop may be a consequence of the decline in vehicle traffic as workplaces and recreational venues began to close.

In contrast, other substances only declined following the introduction of emergency measures. Differences in change patterns are likely attributable to the different sources of pollutants.

Notably, changes are most pronounced for particulate matter, carbon monoxide and nitrogen dioxide, substances related to transportation.

No significant change was observed for sulfur dioxide, which is more closely associated with coal and oil burning for power generation.

These pronounced changes show the profound impact of human activity on the capital’s air. The restrictions imposed in response to COVID-19 are by necessity severe, but also temporary. As life returns to normal and the crisis abates, policymakers may reflect on these changes when considering how to tackle air pollution.


Note: The data and replication code for the analysis presented above is available here. The data analysis used an interrupted time series design, a quasi-experimental method which tests for significance in difference between points in a time series before and after a cut-off. 

Monday, March 23, 2020

Know English and how to use a computer?

A slightly jeering expression in Georgia when speaking about employment prospects suggests that to get a job, you need to know English and how to use computers. Data from Caucasus Barometer 2019 shows there’s a bit of truth in the jest.

Overall, 40% of people on the survey reported having a job. A logistic regression including basic demographic variables, like settlement type, age, gender, minority status and education, suggests that people between the ages of 35 and 54, men, and those with higher education have higher chances of being employed, controlling for other factors. Other demographic factors do not show statistically significant differences.



Aside from the above demographic characteristics, knowledge of English and of computers was also looked at to test the anecdote.  People who report knowing English at a basic level or higher are eight percentage points more likely to be employed, controlling for other factors. Knowing how to use a computer at even a basic level has an even larger effect for 19 percentage points, controlling for other factors. The social and demographic characteristics described above remain significant after controlling for knowledge of English.



Note: Two different logistic regression models were used to generate chart above: (a) self-reported employment in relation to the knowledge of English and (b) self-reported employment in relation to the knowledge of computer. The knowledge question were recoded. Options:  “Beginner”, “Intermediate”, and “Advanced” were coded as ”Beginner or higher”. “No basic knowledge” stayed the same. The regression model for both cases also included the following demographic co-variates: age; gender; ethnicity, education and settlement type. 

In general, these findings align with perceptions of what factors are most important for getting a good job in Georgia. People name education as one of the most important factors for getting a good job in Georgia.

Age, sex, knowledge of English and how to use a computer, and education, are associated with employment in Georgia. This confirms the anecdotal evidence. However knowledge of using a computer in comparison to the knowledge of English appears to be a more important factor for getting a job in Georgia.

To explore more the Caucasus Barometer 2019 survey findings for Georgia, visit CRRC’s Online Data Analysis portal. Replication code for the data analysis is available at CRRC’s GitHub repository here.

Monday, March 16, 2020

Trust in institutions continues its steady decline in Georgia

[Note: This article was co-published with OC Media. The article was written by Dustin Gilbreath, Deputy Research Director at CRRC Georgia. The views presented in the article do not necessarily reflect the views of CRRC-Georgia or any related entity.]

Trust in institutions has been on the decline in Georgia for a decade now. For instance, the level of trust in religious institutions declined from 86% of the public reporting trust in 2008 to 71% in 2019, with the decline being particularly prominent among Orthodox Christians, the main religious group in the country.

Although there has been a decline in trust in most institutions, the decline has been starkest when it comes to political institutions. Newly released data from the Caucasus Barometer 2019 suggests this decline has continued, with the largest decline surrounding trust in the President, Salome Zourabishvili.

Between 2017 and 2019, there were no major increases in trust in institutions. Concomitantly, there were five declines in trust beyond the margin of error. The largest decline in trust was in the president — a 21 percentage point drop.

This likely reflects the change over from Giorgi Margvelashvili to Salome Zourabishvili as president in 2018. Zourabishvili gained office in a heated presidential election, the quality of which was problematic according to some observers. Moreover, few approve of her performance: only 12% reported they viewed her performance positively in a July 2019 NDI and CRRC survey. Hence, the decline is in some senses unsurprising.




Note: Trust in institutions was measured using the question, ‘I will read out a list of social institutions and political unions. Please assess your level of trust toward each of them on a 5-point scale, where code “1” means “Fully distrust”, and “5” means “Fully trust”. First, please tell me how much do you trust or distrust Georgia’s [Institution]?’ Responses of ‘fully trust’ and ‘trust’ are combined into trust for the purposes of this writing.

Aside from the president, there were also declines in trust in parliament and the courts of seven percentage points. There were also declines in trust in the healthcare system and executive government of six and five points, respectively.

While this shows that in the short term, there has been a decline in trust in institutions, there are also notable mid-term trends when it comes to trust in political institutions.

There was a high point in trust in the parliament and executive government in 2012, when the Caucasus Barometer survey took place shortly after the parliamentary elections which unseated the United National Movement.

Similarly, there was an increase in trust in the presidency after the first wave of the Caucasus Barometer survey after Giorgi Margvelashvili was elected (The 2013 wave took place during the presidential election). However, these trends have since reversed, reflecting the growing dissatisfaction with the state of affairs in the country.


While there have been several mid-term trends in the data, taking the long view suggests that the drop in trust is a longer-term phenomenon that is larger than the administration of one political party or another.

Trust in all the domestic institutions asked about on the Caucasus Barometer survey have declined since the question was first asked, with the exception of the Army.

Not only has trust declined for all institutions aside from the Army, it has generally done so by large amounts. Compared with 2008, trust in the President has declined by 35 percentage points, in the media by 30 points, and in the Public Defender’s Office and banks by 29 and 28 points respectively. Even the highly trusted police force has experienced a five-point decline in the share of the population reporting they trust them since 2008. The average decline in trust was 17 percentage points between 2008 and 2019.


Note: Trust in political parties was first measured in 2012. Trust in local government was first measured in 2009. All other institutions were first measured in 2008.

The drops in trust in institutions in Georgia are not the only sign that all is not well. Data from other sources suggest that people increasingly think Georgia is heading in the wrong direction.  Fewer people are optimistic about the state of Georgia. Fewer people are satisfied with life. There have been large drops in the belief that most people can be trusted in Georgia. Taken together, the above points to stagnation in Georgia.

The data used in this blog post is available here.



Monday, March 09, 2020

What kind of electoral system do Georgians actually want?

[Note: This article was written by David Sichinava, Research Director at CRRC Georgia. The views presented in the article do not necessarily represent the views of CRRC Georgia or any related entity. The article was co-published with OC-Media.]


On 8 March, Georgia’s political leaders agreed on a new electoral system under which 120 seats will be allocated via proportional elections and 30 seats will be allocated via direct election of candidates.
The long-fought-over electoral reform was a compromise which represents two steps forward after three steps had been taken back.

The debate over Georgia’s electoral system fueled last year’s political crisis. As the parliament ditched a promised constitutional amendment instituting fully proportional elections for the legislature instead of the current mixed system, opposition parties and civil society groups hit the streets of Tbilisi to protest.

EU-mediated talks between the government and opposition also stalled after the arrest of prominent opposition leader Gigi Ugulava, and disagreement over the interim model of elections.

While both the ruling Georgian Dream Party and the opposition argued that their own initiatives were publicly popular, a recent CRRC Georgia survey shows that the public is more ambivalent than might be expected and sometimes inchoate in their views.

While the majority of those aware that Georgian Dream buried the constitutional amendments assess this negatively, Georgians are split when it comes to potential models for the electoral system.
CRRC Georgia’s omnibus survey, which was fielded in mid-January, showed that over three-quarters of Georgians (76%) were well aware that the ruling party did not pass constitutional amendments.
Most (60%) who had heard of the failure to pass the amendments disapproved of the decision. About a fifth (23%) of Georgians approved of parliament’s decision not to vote in favour of the amendments, while others did not know what to say.

Attitudes vary by partisanship. Almost half of the Georgian Dream supporters (44%) that were aware of the failure to pass the legislation view the failure negatively.

In contrast, those who support opposition parties overwhelmingly disapprove of Georgian Dream’s failure to pass the amendment.


Note: Party identification was coded as follows: supporters of the UNM, European Georgia, Lelo, Civic Movement, Girchi, and For New Georgia were categorised as ‘Liberals’. Supporters of the Alliance of Patriots of Georgia, the Democratic Movement, and Labour Party were grouped in ‘other’.

While it is now clear what electoral system the 2020 parliamentary elections will be conducted under, what did the public want? Suggested proposals for electing MPs to parliament were confusing and potentially hard to grasp for the general public.

The initial proposal suggested and then ditched by Georgian Dream was to have a fully proportional system without any electoral threshold.

Later, the governing party proposed retaining the current mixed electoral system while the opposition supported a ‘German model’, which leaves the mixed model of representation but allocates additional seats to reflect the popular vote.

To avoid such confusion, respondents were separately asked whether they supported specific components of these systems. First, respondents choose between whether they preferred voting for only majoritarian candidates, only for parties, or for both.

Next, those who supported either voting for parties alone or a mixed system were asked whether they approved of having a threshold. Finally, respondents who preferred a mixed electoral system were asked whether they preferred allocation of seats proportional to the popular vote.

Overall, the survey suggests that a plurality (47%) of Georgians support a mixed model of elections to parliament where citizens vote for both parties and individual candidates.

Only 14% support party-list voting only, and 15% think that Georgians should only vote for specific candidates.

About a quarter of Georgians have ambivalent feelings: 14% say that it does not matter to which model Georgia sticks to, while 11% say that they don’t know.

Attitudes are similar across party lines with the exception of liberal opposition parties. Among their supporters, 38% think that Georgians should vote for candidates only, while 29% prefer to vote for parties.

In the initial draft of amendments, Georgian Dream proposed getting rid of the electoral threshold in order to ensure representation of relatively minor political groups. According to the survey, those who prefer a proportional (14%) or mixed model (47%) of electing MPs, overwhelmingly (85%) support retaining one.

Supporters of a mixed model of electing MPs tend to support a proportional distribution of seats. Around half of the supporters of a mixed model (51%) support a party-list vote or a mixed system where seats are assigned proportionally to the popular vote.

Around 27% support either a majoritarian system or the current, mixed system.

Thirty-six per cent of Georgians are ambivalent — they either do not have a preferred way of electing MPs, do not know, or refuse to answer on these questions.



Note: Support for different models of allocating mandates is the combination of two different questions. Respondents who support fully proportional representation and those who are for a mixed system and proportional allocation of mandates within mixed representation are grouped into the proportional allocation group. Those supporting full majoritarian representation fall into a separate category. Respondents who prefer a mixed system and do not favour distribution of mandates per the vote share are grouped together. Those who do not have any preference in terms of seat allocation (‘Don’t know’, ‘Refuse to answer’, and ‘Does not matter’) are put into the ambivalent category.

Importantly, much of the public is inchoate in their attitudes towards the electoral systems. Overall, those who found it unacceptable to ditch the amendments are more likely to support a system which ensures proportional allocation of seats than those who were fine with the failure to pass the electoral reform (46% versus 37%).

Yet, 27% of the public who thought that it was unacceptable to ditch the constitutional amendments also report that they do not support the proportional allocation of seats.



The key challenge for Georgia’s electoral system is whether it can represent political parties proportional to their popular support. Neither the current mixed system nor a majoritarian system necessarily yields proportional allocation of seats.

Indeed, in the past, proportional representation has almost never happened, with the exception of the highly contested and polarised 2012 parliamentary elections.

Importantly, there is a relative consensus among Georgians that have partisan sympathies. A plurality of Georgian Dream supporters (42%), and the majority of those sympathising with both the liberal (53%) and conservative (56%) opposition parties prefer models ensuring proportional representation.
Those who are not affiliated with any political party, declined to disclose their preferences, or do not know are more likely to be ambivalent.

In short, the preference for a mixed model of parliamentary elections prevails among Georgians. This suggests that voters in Georgia may want to see at least some politicians with connections to their communities in parliament.

Still, a plurality of partisan voters prefer an electoral system yielding proportional representation, both in the governmental and opposition camps.

The findings of CRRC Georgia’s omnibus survey substantiates the argument that there is a considerable consensus across party lines for having a variant of a mixed system where the final tally of seats is assigned relative to the popular vote.

Electoral rules often reflect compromises made by political groups. The current opinion of Georgians also hints that Georgians would prefer a balance be struck between more radical proposals.
The dataset, the questionnaire, and the replication code used in the above article can be found here.

To find out more about CRRC Georgia’s omnibus surveys, click here.

Monday, March 02, 2020

How widespread is homophobia in Georgia?

[Note: This piece was co-published with OC Media. This article was written by Dustin Gilbreath, Deputy Research Director at CRRC Georgia. The views presented in this article do not represent the views of UN Women, CRRC Georgia, or any related entity.]

Homophobia is widespread in Georgia. The homophobic riots that occurred on the International Day against Homophobia in 2013 and the bedlam that took place surrounding the planning of the 2019 Pride Parade exemplify this.

The 2019 Caucasus Barometer survey asked two questions proxying homophobia. The first asked whether or not people would approve of someone like them doing business with a homosexual. The second asked people to name the group that they would least like to have as neighbours from a list of different groups, including homosexuals, criminals, people following a different religion, people with different political views, Europeans or Asians who come to live in Georgia and want to stay, and drug addicts.

Nearly nine in ten people (87%) would disapprove of a person like them doing business with a homosexual. By comparison 24% of the public reports they would least like a homosexual as a neighbour among the groups asked about.

Aside from asking about doing business with homosexuals, the survey asked the same question about 18 ethnic and religious groups. More people approved of doing business with every other group asked about on the survey. Indeed, the only other group that people are remotely as negative about are Jehovah’s Witnesses who 81% of people disapprove of people like them doing business with.

A majority of the public approves of people of their ethnicity doing business with all the other groups asked about on the survey.  The average share of people that approve of groups aside from homosexuals is 65%.

The above data lead to the question of who is more or less likely to be tolerant towards homosexuals? The data suggest that people in Tbilisi, ethnic Georgians, those with more education, women, people in wealthier households, and younger people are all more likely to approve of someone like them doing business with a homosexual.

There is not a significant difference between people who are employed or not or use the internet more or less often. The charts below show the differences controlling for these factors.



The question about neighbours shows a slightly different picture. On this question, there are significant differences between ethnic minorities and ethnic Georgians, men and women, and between settlement types.

People outside Tbilisi are seven percentage points more likely to name homosexuals as the group they would least like to have as neighbours. Ethnic minorities are 12 percentage points less likely to name homosexuals than ethnic Georgians. Women are 11 percentage points less likely to name homosexuals than men. The remaining variables tested showed no significant differences.

The above shows that homophobia is relatively widespread in Georgia. Men are more likely to be homophobic than women. So are ethnic minorities more than ethnic Georgians. Younger people, those with more education, and wealthier people express homophobic attitudes less often though still frequently.

The data used for the analysis presented in the article is available here. Replication code for the data analysis is available here.

The data analysis presented in this article made use of a logistic regression. The outcome variables were whether or not an individual approved of someone like them doing business with a homosexual and whether or not they named homosexuals as their least desired neighbour. The independent variables included sex (male or female), age, settlement type (capital, other urban, and rural), employment status (working or not), years in formal education, ethnicity (minority or not), internet usage (daily user or not), and wealth proxied by the number of assets a family-owned from a list of 10 possible assets.