COVID-19 and Climate Shocks: Gendered effects on political preferences & behavior

We have developed a survey instrument that will be deployed next month via short phone surveys. These phone surveys will be conducted with a sample 3,000 respondents from 1,500 households, balanced by gender and residence in one of three divisions with varied exposure to climate shocks and Covid-19: Rangpur, Dhaka, or Chittagong. We are also collecting a panel of high-frequency local geospatial data on weather patterns that will enable objective identification of climate shocks at a micro-level, which we will merge with self-collected data on unique forms of gendered preferences, intra-household behavior, engagement with the state, and collective action.

2020-10-05

2021-05-04

Preferences for systemic change, support for political (collective) action to improve welfare (personally and by others), willingness to support daughters and sons' economic advancement, self-reported political engagement.

We hypothesize that climate- or covid-shocks affect women's political preferences and behavior by altering the following:
1. Magnitude of gendered grievances: Relative to pre-crisis levels, the more adverse the effect on aspects of household or economic life for which women are mainly responsible, the more likely they are to prefer gendered redistribution of resources. This should encourage participation in politics of some form, as well as a preference for political action as a means to improve societal well-being.
2. Domain of women's agency: The more aspects of household and economic life over which women's agency extends, the more likely it is that they will have grievances that necessitate political engagement. Chiefly, shocks can alter women's domain of agency by changing (i) the proportion of household income earned by women, or (ii) women's involvement in household decision-making.
3. Women's capacity for political engagement: Shocks that reduce (increase) women's mobility and extra-household networks will reduce (increase) their political engagement. Further, we hypothesize that these shocks disproportionately increase the risks and
responsibilities women bear (relative to men) for ensuring personal and familial survival.

The testable hypotheses that follow from the above are as follows:
H1. Both climate shocks and COVID-19 increase women's preference (relative to men's) for systemic change brought about by means of political action.
H2. Climate shocks (ceteris paribus) increase women's political participation on average, by increasing all of 1, 2 and 3 above.
H3. COVID-19 (ceteris paribus) decreases women's political participation on average. COVID-19 increases 1, but decreases 2 and 3, with the resulting net effect being negative.
Note that overall, the combined effect of climate shocks and COVID-19 on women's political participation is theoretically ambiguous. Our expectation is that, where both shocks apply, the relative magnitude of each shock's impact on the scope of gendered agency will predict which shock drives gendered political preferences and behavior.

Intra-household bargaining, demographics, participation in varied (proximate versus distant) economic activities, and resulting individual-level financial contributions to the household.

The dynamic underlying these hypotheses is that these crises affect political outcomes based on how they alter economic livelihoods, as well as gendered patterns of migration and economic opportunity. Specifically, we expect that climate change-induced disasters are likely to induce male out-migration (from rural to urban areas) and pressure both men and women to generate new income. If so, female economic (and social) contributions to the household should increase alongside their financial autonomy post-crises, opening new opportunities for women's political engagement and an impetus to rethink traditional (political, social, and economic) preferences.
In contrast, we expect the COVID-19 crisis increases reverse migration (men returning from urban to rural areas) and reduces economic opportunities, particularly for women. In this case, female economic contributions and autonomy should diminish in the wake of COVID-19, reducing opportunities for women's political engagement and increasing pressure to support traditional preferences. The simultaneous experience of both crises may either magnify both the impetus and capacity for women's political engagement--if, for instance both COVID-19 and climate shocks are perceived as unjustly diminishing women's current economic returns relative to female expectations pre-crisis, and these shocks simultaneously expand demands for women to contribute more to their households, broadening their agency to collectively reshape future opportunities--or, where COVID- 19 is most severe, rigidly-enforced lockdowns may dominate household calculations and thus magnify the restrictions on women's political engagement.

We embed experiments within our phone surveys.

The survey includes sections on how COVID-19 and recent natural hazards (if any) have affected respondents' lives, and an experiment designed to test H1. While discussing the wage benefits and COVID-related risks associated with employment in the garment manufacturing sector, the Control condition provides optimism about better protection of "some lucky workers" by their employers, including "providing higher, more stable wages and better working conditions". In contrast, Treatment 1 ties these protections
to political activity (protests) by workers, adding, "Where workers have stood up to employers and protested on the streets, as over 50,000 did in the winter of 2018-19, some employers have responded by making changes to protect them, providing higher, more
stable wages and better working conditions". Treatment 2 adds to treatment 1 by introducing a specific gender lens, "Where female workers have stood up to employers and protested on the streets, as over 50,000 did in the winter of 2018-19, some employers have responded by making changes to protect them, providing higher, more stable wages and better working conditions".

Completed by a computer.

Individual

No

3,000 individuals

3,000 individuals

1,000 individuals in control; 1,000 individuals in treatment 1; 1,000 individuals in treatment 2.