Coronavirus, flexibility of work arrangements and financial security

In this survey experiment, survey respondents will be randomized into one of four groups: ‘Control’, ‘Treatment 1’, ‘Treatment 2’, or ‘Treatment 3’. Randomization will be performed at the individual level and each participant will have a 25% chance of being randomized into any given group. Individuals in the ‘Treatment 1’ group will receive truthful information on health-related aspects of the coronavirus outbreak. ‘Treatment 2’ will receive the same information as ‘Treatment 1’ and will additionally be informed about economic consequences of the outbreak. ‘Treatment 3’ will receive the same information as ‘Treatment 1’ and will additionally receive information on workers without paid sick leave. The ‘Control’ group will receive none of the above information.

Control:
As you may have heard, we are experiencing an outbreak of respiratory illness, COVID-19, caused by a new coronavirus. To help stop the spread of the virus, various social-distancing measures are being put in place.

Treatment 1:
As you may have heard, we are experiencing an outbreak of respiratory illness, COVID-19, caused by a new coronavirus. To help stop the spread of the virus, various social-distancing measures are being put in place. In different countries, officials predict that more than 70% of people might get infected. While most people will only develop mild symptoms, the virus can be severe for older people, many of whom may require hospital treatment. This has already put a lot of pressure on the health systems in countries where the outbreak started earlier. For each age group, the chart below shows the estimated proportion of coronavirus cases with symptoms that need hospital treatment. Source: Imperial College COVID-19 Response Team. <insert chart>

Treatment 2:
As you may have heard, we are experiencing an outbreak of respiratory illness, COVID-19, caused by a new coronavirus. To help stop the spread of the virus, various social-distancing measures are being put in place. In different countries, officials predict that more than 70% of people might get infected. While most people will only develop mild symptoms, the virus can be severe for older people, many of whom may require hospital treatment. This has already put a lot of pressure on the health systems in countries where the outbreak started earlier. For each age group, the chart below shows the estimated proportion of coronavirus cases with symptoms that need hospital treatment. Source: Imperial College COVID-19 Response Team. <insert chart>

The virus is predicted to have a big impact on the whole economy. In the United States, unemployment has already risen sharply. 281,000 people became unemployed in the week ending 14 March, a sharp rise from 211,000 in the previous week. This rise is larger than any week-to-week unemployment movement during (or since) the 2008 financial crisis. Many businesses have already been affected by a fall in revenue caused by social-distancing measures. Anne*, a small business owner, has seen many orders cancelled. With no cash coming in, she says she was forced to lay off most of her 17 employees. * Not her real name.

Treatment 3:
As you may have heard, we are experiencing an outbreak of respiratory illness, COVID-19, caused by a new coronavirus. To help stop the spread of the virus, various social-distancing measures are being put in place. In different countries, officials predict that more than 70% of people might get infected. While most people will only develop mild symptoms, the virus can be severe for older people, many of whom may require hospital treatment. This has already put a lot of pressure on the health systems in countries where the outbreak started earlier. For each age group, the chart below shows the estimated proportion of coronavirus cases with symptoms that need hospital treatment. Source: Imperial College COVID-19 Response Team. <insert chart>

Many people are not entitled to paid sick leave. This puts them in a difficult situation if they risk losing their job or their income if they stay home. Adam*, who is self-employed, said in response to the outbreak: “If you’re self-employed you have to continue working. I’m not about to make my children starve because of coronavirus. If I’m physically able to work, then isolation is not happening for me.” Adam admits that continuing to work might spread the virus. “That’s a risk I would have to take,” he said. Not granting paid sick leave to all workers poses serious threats to public health. *Not his real name.

2020-03-25

2020-05-25

We will estimate the impact of the treatments on several outcomes, which can be categorized into three different groups: (1) Individual perceptions of the severity of the coronavirus outbreak and the adequacy of the policy response to the crisis, (2) individual valuations of different work arrangements, and (3) individual support for public policies.

(1) Perceptions about the severity of the coronavirus outbreak and adequacy of policy response to the crisis:
First, respondents are asked how the coronavirus outbreak has already affected their daily lives. Second, we ask for respondents’ subjective likelihood that certain outcomes will occur in the near future. Outcomes include disruptions to labor supply (e.g. having to work fewer hours, having to shut the business or having to rearrange work schedule to care for others) and loss in earnings, as well as estimates of the magnitude of the outbreak in the respondents’ state or region, measured as the perceived percentage of people who will contract the coronavirus in their area.

(2) Individual valuations of different work arrangements:
We follow Mas and Pallais (2017) and elicit individual willingness to pay or willingness to accept for different work attributes by using a simple discrete choice experiment. The work arrangements we consider are entitlement to sick pay, flexible scheduling and a permanent versus zero-hour (or on-call) contract. For each arrangement, respondents who currently have a paid job are presented with the choice between their current work contract and a new contract that excludes (includes) the arrangement under analysis and pays more (less) per hour. Unemployed individuals are offered the choice between two contracts that are similar to their last job contract in all aspects but the arrangement under analysis and wage. We randomly vary the wage difference between the two options and ask respondents which arrangement they would prefer. We implement a between-subject design where each respondent is presented with two options for each job characteristic under analysis, to limit cognitive load.

(3) Individual support for public policies:
Respondents are asked Likert-scale questions on the extent to which they agree or disagree with statements about public policies in support of vulnerable groups. Policies will include measures already taken by the government to tackle the outbreak, as well as proposals to provide government support to self-employed, temporary workers and low-income citizens. We will also use similar Likert-scale questions to measure respondents’ perception about the readiness of the government and the health system to tackle the coronavirus outbreak. Finally, we will measure support for redistributive policies targeted mostly at families or businesses.

Within each country, we will survey 4,000 individuals who are part of an online panel. Individuals will be randomized into one of four groups: ‘Control’, ‘Treatment 1’, ‘Treatment 2’, or ‘Treatment 3’. Randomization will be performed at the individual level and each participant will have a 25% chance of being randomized into any given group. We will estimate the causal impact of the treatments on a range of different outcomes and analyse whether the treatments were more/less effective for different groups in the population. This survey experiment will help us understand how individuals respond to the different treatments.

We will estimate the average treatment effect of the treatments on each outcome by running OLS regressions of the outcome on the treatment dummies. To increase precision of our estimates, we will also perform the analyses controlling for baseline characteristics. In addition, we will look at several dimensions of heterogeneity. More specifically, we are interested in whether the treatments are differentially effective for (i) democrats vs republicans, (ii) men vs women, (iii) above vs below median age, and (iv) people with different work arrangements at baseline.

Randomization done by the survey software we use to conduct the survey.

Individual

No

The randomization is performed at the individual level. We do not have any clusters.

4,000 individuals per country

Approximately 1,000 individuals ‘Control’, 1,000 individuals ‘Treatment 1’, 1,000 individuals ‘Treatment 2’, 1,000 individuals ‘Treatment 3’