Do movement restrictions affect the well-being of the population?

Responsible researcher: Eduarda Miller de Figueiredo

Article title: COVID-19, LOCKDOWNS AND WELL-BEING:
EVIDENCE FROM GOOGLE TRENDS

Article authors: Abel Brodeur, Andrew E. Clark, SarahFleche and Nattavudh Powdthavee

Sample Size: Varies by search term

Location of intervention : European countries and the United States

Sector : Health

Type of intervention : Effects of movement restrictions on population well-being

Variable of main interest : boredom, contentment, divorce, commitment, irritability, loneliness,
panic, sadness, sleep, stress, suicide, well-being and worry

Assessment method : Differences-in-differences

Policy Problem

The COVID-19 pandemic required a rapid response from all countries to save as many lives as possible in the short and medium term. To this end, most European countries and the United States imposed lockdowns on their residents, in accordance with epidemiological model guidelines to contain the spread of the virus (Ferguson et al., 2020).

These movement restrictions have effects on GDP, levels of trust in governments, education and the well-being of the population. The by-products of lockdown involve unemployment, social isolation and lack of freedom, which are risk factors for mental health and unhappiness (Leigh-Hunt et al., 2017).

Assessment Context

There is some research underway regarding the evolution of the population's well-being during the pandemic. However, to fully assess this effect there is a need for data prior to the pandemic and lockdown. In most existing research, such data are unavailable.

To overcome this data problem, the authors analyze Google Trends data between January 1, 2009 and April 10 in countries that introduced a full lockdown at the end of the period. Whereas, Google search signals provide accurate and representative information about users' current search behavior and sentiment. Additionally, Google Trends shows aggregated measures of search activity in a location and is therefore less vulnerable to small sample bias (Baker and Fradkin, 2017).

In this way, the article contributed to the literature by documenting the impacts of social restrictions on the mental health of the population.

Policy Details

Google Trends data provides an unfiltered sample of searches performed on Google. A search term query in Google Trends returns searches for an exact search term, while a topic query includes search related to terms regardless of the idiot.

Therefore, it provides an index for the search intensity for a topic or search term over the period in question and in a requested geographic area. This index ranges from 0 to 100, where 100 is the day with the most searches for the topic and 0 indicates that a given day had no search volume for the specific term.

In this research, the authors used the following search terms for topics related to well-being between January 1, 2019 and April 10, 2020: boredom, contentment, divorce, commitment, irritability, loneliness, panic, sadness, sleep, stress, suicide, well-being and worry. These topics come from different items of the General Health Questionnaire (GHQ)[1].

Therefore, the authors have a database of these topics for countries that introduced a lockdown at the end of the period considered, namely: Austria, Belgium, France, Ireland, Italy, Luxembourg, Portugal, Spain, United Kingdom and United States.

Methodology Details

The 2019 daily data were obtained in a separate request from the 2020 daily data. Therefore, the authors needed to resize the two series to the same scale factor of the score from 0-100, thus being able to compare them.

To this end, the respective weekly research interest weights were first calculated for all weeks of the period, aggregating to calculate the weekly average of searches for the topic in country c . From this, the daily data was resized for each separate period by multiplying the 2019 weekly average by the 2019 weekly search interest weight, repeating the process for the 2020 period. And, finally, the values ​​were normalized between 0 and 100 .

Assuming that, in the absence of lockdown, Google users' behaviors would have evolved in the same way as in the year before the lockdown, the authors use a Differences-in-Differences (DiD) estimator to estimate the joint effect of the Covid-19 pandemic and lockdowns with research related to well-being. In this way, comparing pre- and post-lockdown surveys in 2020 to pre- and post-lockdown surveys from the same period in 2019, thus ensuring that countries' seasonal changes were not behind the findings. As the psychological effects of the lockdown may have started from the moment the policy was announced to the public, the authors consider the “lockdown date” to be the date on which the restriction was announced.

In the DiD model, the authors used Google search topics related to well-being as a dependent variable, including fixed effects of country, state, year, week and day. Furthermore, the lagged number of new Covid-19 deaths per day per million in the country or state was also controlled.

To test the immediate structural break caused by the lockdown, the study also performs a Regression Discontinuity (RDD), to identify potential breaks in two series – pre and post lockdown. Where the dependent variable is the absolute distance in days from the announcement of the “stay at home” order: negative for the previous days and positive for the subsequent days. Thus, the actual or counterfactual announcement date is defined as day zero.

Results

Searches for “boredom” in Europe saw a sharp increase around the date of the announcement in 2020, while in the United States, which began lockdown later, this search began 10 days before the announcement. This pattern was only observed in 2020, with no sudden changes in the same period in 2019.

It was also possible to observe a notable increase after the lockdown in the search for “solitude” in Europe, which was not observed in the United States. On the other hand, both had an increase in searches for “sadness” around one to two weeks after the lockdown.

The Differences-in-Differences estimation demonstrates that the lockdown variable produced a significant increase in the search intensity for “boredom” in both locations, this increase being significant at 1%. A significant increase in searches for “loneliness”, “worry” and “sadness” was also observed.

Another result observed was the statistically significant drops in “stress”, “suicide” and “divorce” in both locations. However, no effect on “sleep” was found in European countries. Regarding the theme “well-being”, the results differ between locations. In the USA there was a positive effect on the intensity of research related to the topic, but in Europe it had a negative effect.

When the authors divided Europe into early and late lockdowns, with the late lockdown comprising Ireland, Portugal and the United Kingdom, they found a positive effect on well-being relative to the late group. Thus, they observed that the effect of lockdown on well-being measures is often more positive in countries with a late lockdown. Thus, those who entered late lockdowns may be less stressed, but public health benefits were seen more strongly in countries that entered early lockdowns.

The results of the Regression Discontinuity (RDD) demonstrated that the immediate effect of the lockdown is an increase in searches for “boredom” and “commitment” and a reduction in searches for “panic”. However, there was little short-term impact on “stress”, “sadness”, “suicide” and “worry”.

Public Policy Lessons

Despite the need expressed by governments that society needs to stay at home to save lives, evidence suggests that people's mental health has been affected during the first weeks of lockdown. Therefore, the need to emphasize the benefits of lockdown for the health of society is clear, making sure that there is appropriate support to help those who struggle most with the lockdown which, according to Oswald and Powdthavee (2020), starts with the younger generations .

Reference

BRODEUR, Abel et al. COVID-19, lockdowns and well-being: Evidence from Google Trends. Journal of public economics, vol. 193, p. 104346, 2021.

[1] General Health Questionnaire (GHQ).